# StandardDB - Full Content

> This file contains the complete content of StandardDB, a changelog aggregation platform tracking 109 companies, 31 products, and 395 changelog entries across the AI and developer tools ecosystem.

## Companies

### AI21 Labs
- Website: https://docs.ai21.com
- Description: AI21 Labs — Frontier AI Lab
- StandardDB page: https://standarddb.com/ai21

### Aider
- Website: https://aider.chat
- Description: Aider — AI Code Editor
- StandardDB page: https://standarddb.com/aider

### Alibaba (Qwen)
- Website: https://qwenlm.github.io
- Description: Alibaba (Qwen) — Frontier AI Lab
- StandardDB page: https://standarddb.com/qwen

### Allen Institute for AI
- Website: https://allenai.org
- Description: Allen Institute for AI — Open Source / Research
- StandardDB page: https://standarddb.com/ai2

### Anthropic
- Website: https://anthropic.com
- Description: Anthropic is an AI safety company building reliable, interpretable, and steerable AI systems. Makers of Claude.
- StandardDB page: https://standarddb.com/anthropic

### Anthropic (MCP)
- Website: https://modelcontextprotocol.io
- Description: Anthropic (MCP) — Protocol
- StandardDB page: https://standarddb.com/mcp

### Apple
- Website: https://machinelearning.apple.com
- Description: Apple — Hardware
- StandardDB page: https://standarddb.com/apple

### Artie
- Website: https://www.artie.com
- Description: Artie is a real-time data replication platform that streams data from databases to data warehouses using change data capture.
- StandardDB page: https://standarddb.com/artie

### AssemblyAI
- Website: https://www.assemblyai.com
- Description: AssemblyAI — AI Voice
- StandardDB page: https://standarddb.com/assemblyai

### Augment Code
- Website: https://docs.augmentcode.com
- Description: Augment Code — AI Code Editor
- StandardDB page: https://standarddb.com/augment

### Braintrust
- Website: https://www.braintrust.dev
- Description: Braintrust — Data & Observability
- StandardDB page: https://standarddb.com/braintrust

### Browserbase
- Website: https://docs.browserbase.com
- Description: Browserbase — Agents & Automation
- StandardDB page: https://standarddb.com/browserbase

### Bun
- Website: https://bun.sh
- Description: Bun — Developer Platform
- StandardDB page: https://standarddb.com/bun

### Canva
- Website: https://www.canva.com
- Description: Canva — Design
- StandardDB page: https://standarddb.com/canva

### Character.AI
- Website: https://blog.character.ai
- Description: Character.AI — Social / Consumer
- StandardDB page: https://standarddb.com/character

### Cloudflare
- Website: https://cloudflare.com
- Description: Cloudflare is a global cloud platform that delivers a broad range of network services to businesses of all sizes, making them more secure while enhancing performance and reliability.
- StandardDB page: https://standarddb.com/cloudflare

### Cognition (Devin)
- Website: https://docs.devin.ai
- Description: Cognition (Devin) — AI Code Editor
- StandardDB page: https://standarddb.com/devin

### Cohere
- Website: https://docs.cohere.com
- Description: Cohere — Frontier AI Lab
- StandardDB page: https://standarddb.com/cohere

### Composio
- Website: https://docs.composio.dev
- Description: Composio — Agents & Automation
- StandardDB page: https://standarddb.com/composio

### Concourse
- Website: https://www.concourse.co
- Description: Concourse is an AI platform that provides AI agents for corporate finance teams to automate tasks, deliver insights, and create financial reports.
- StandardDB page: https://standarddb.com/concourse

### Continue
- Website: https://github.com
- Description: Continue — AI Code Editor
- StandardDB page: https://standarddb.com/continue

### Cursor
- Website: https://www.cursor.com
- Description: Cursor — AI Code Editor
- StandardDB page: https://standarddb.com/cursor

### Databricks
- Website: https://docs.databricks.com
- Description: Databricks — AI Infrastructure
- StandardDB page: https://standarddb.com/databricks

### Deepgram
- Website: https://developers.deepgram.com
- Description: Deepgram — AI Voice
- StandardDB page: https://standarddb.com/deepgram

### DeepSeek
- Website: https://api-docs.deepseek.com
- Description: DeepSeek — Frontier AI Lab
- StandardDB page: https://standarddb.com/deepseek

### Deno
- Website: https://deno.com
- Description: Deno — Developer Platform
- StandardDB page: https://standarddb.com/deno

### Descript
- Website: https://www.descript.com
- Description: Descript — AI Creative
- StandardDB page: https://standarddb.com/descript

### Discord
- Website: https://discord.com
- Description: Discord — Social / Consumer
- StandardDB page: https://standarddb.com/discord

### Docker
- Website: https://docs.docker.com
- Description: Docker — Developer Platform
- StandardDB page: https://standarddb.com/docker

### Duolingo
- Website: https://blog.duolingo.com
- Description: Duolingo — Social / Consumer
- StandardDB page: https://standarddb.com/duolingo

### E2B
- Website: https://e2b.dev
- Description: E2B — Agents & Automation
- StandardDB page: https://standarddb.com/e2b

### ElevenLabs
- Website: https://elevenlabs.io
- Description: ElevenLabs — AI Creative
- StandardDB page: https://standarddb.com/elevenlabs

### Expo
- Website: https://expo.dev
- Description: Expo — Developer Platform
- StandardDB page: https://standarddb.com/expo

### Figma
- Website: https://www.figma.com
- Description: Figma — Design
- StandardDB page: https://standarddb.com/figma

### Fireworks AI
- Website: https://docs.fireworks.ai
- Description: Fireworks AI — AI Infrastructure
- StandardDB page: https://standarddb.com/fireworks

### Fly.io
- Website: https://fly.io
- Description: Fly.io — Developer Platform
- StandardDB page: https://standarddb.com/fly

### Framer
- Website: https://www.framer.com
- Description: Framer — Design
- StandardDB page: https://standarddb.com/framer

### Gamma
- Website: https://gamma.app
- Description: Gamma — Productivity
- StandardDB page: https://standarddb.com/gamma

### GitHub
- Website: https://github.com
- Description: GitHub is the world's leading AI-powered developer platform for building, scaling, and delivering secure software.
- StandardDB page: https://standarddb.com/github

### Glean
- Website: https://www.glean.com
- Description: Glean — AI Search
- StandardDB page: https://standarddb.com/glean

### Google
- Website: https://google.com
- Description: Google builds products and platforms that help people and businesses. Their AI division develops Gemini and other frontier models.
- StandardDB page: https://standarddb.com/google

### Grammarly
- Website: https://support.grammarly.com
- Description: Grammarly — Productivity
- StandardDB page: https://standarddb.com/grammarly

### Granola
- Website: https://www.granola.ai
- Description: Granola — Productivity
- StandardDB page: https://standarddb.com/granola

### Groq
- Website: https://console.groq.com
- Description: Groq — AI Infrastructure
- StandardDB page: https://standarddb.com/groq

### Harvey AI
- Website: https://www.harvey.ai
- Description: Harvey AI — Fintech / Legal
- StandardDB page: https://standarddb.com/harvey

### Helicone
- Website: https://helicone.ai
- Description: Helicone — Data & Observability
- StandardDB page: https://standarddb.com/helicone

### Hex
- Website: https://hex.tech
- Description: Hex — Data & Observability
- StandardDB page: https://standarddb.com/hex

### HeyGen
- Website: https://www.heygen.com
- Description: HeyGen — AI Creative
- StandardDB page: https://standarddb.com/heygen

### Hugging Face
- Website: https://huggingface.co
- Description: Hugging Face — AI Infrastructure
- StandardDB page: https://standarddb.com/huggingface

### Ideogram
- Website: https://about.ideogram.ai
- Description: Ideogram — AI Creative
- StandardDB page: https://standarddb.com/ideogram

### Intercom
- Website: https://www.intercom.com
- Description: Intercom — Customer / Support
- StandardDB page: https://standarddb.com/intercom

### JetBrains
- Website: https://www.jetbrains.com
- Description: JetBrains — AI Code Editor
- StandardDB page: https://standarddb.com/jetbrains

### Kagi
- Website: https://kagi.com
- Description: Kagi — AI Search
- StandardDB page: https://standarddb.com/kagi

### LangChain
- Website: https://blog.langchain.dev
- Description: LangChain — AI Infrastructure
- StandardDB page: https://standarddb.com/langchain

### Leonardo AI
- Website: https://leonardo.ai
- Description: Leonardo AI — AI Creative
- StandardDB page: https://standarddb.com/leonardo

### Linear
- Website: https://linear.app
- Description: Linear is a purpose-built tool for planning and building products. Streamline issues, projects, and product roadmaps.
- StandardDB page: https://standarddb.com/linear

### LlamaIndex
- Website: https://github.com
- Description: LlamaIndex — AI Infrastructure
- StandardDB page: https://standarddb.com/llamaindex

### LMSYS
- Website: https://lmsys.org
- Description: LMSYS — Open Source / Research
- StandardDB page: https://standarddb.com/lmsys

### Lovable
- Website: https://lovable.featurebase.app
- Description: Lovable — AI App Builder
- StandardDB page: https://standarddb.com/lovable

### Magic Patterns
- Website: https://www.magicpatterns.com
- Description: Magic Patterns is an AI-powered design tool that generates user interfaces from text prompts, images, or Figma mockups.
- StandardDB page: https://standarddb.com/magic-patterns

### Meta
- Website: https://ai.meta.com
- Description: Meta — Frontier AI Lab
- StandardDB page: https://standarddb.com/meta

### Midjourney
- Website: https://www.midjourney.com
- Description: Midjourney — AI Creative
- StandardDB page: https://standarddb.com/midjourney

### Mistral AI
- Website: https://docs.mistral.ai
- Description: Mistral AI — Frontier AI Lab
- StandardDB page: https://standarddb.com/mistral

### Modal
- Website: https://modal.com
- Description: Modal — AI Infrastructure
- StandardDB page: https://standarddb.com/modal

### n8n
- Website: https://docs.n8n.io
- Description: n8n — Agents & Automation
- StandardDB page: https://standarddb.com/n8n

### Neon
- Website: https://neon.tech
- Description: Neon — Developer Platform
- StandardDB page: https://standarddb.com/neon

### Notion
- Website: https://www.notion.com
- Description: Notion — Developer Platform
- StandardDB page: https://standarddb.com/notion

### NVIDIA
- Website: https://developer.nvidia.com
- Description: NVIDIA — Hardware
- StandardDB page: https://standarddb.com/nvidia

### Ollama
- Website: https://github.com
- Description: Ollama — AI Infrastructure
- StandardDB page: https://standarddb.com/ollama

### OpenAI
- Website: https://openai.com
- Description: OpenAI is an AI research and deployment company. Creators of GPT, DALL-E, and ChatGPT.
- StandardDB page: https://standarddb.com/openai

### Otter.ai
- Website: https://blog.otter.ai
- Description: Otter.ai — Productivity
- StandardDB page: https://standarddb.com/otter

### Perplexity
- Website: https://www.perplexity.ai
- Description: Perplexity — AI Search
- StandardDB page: https://standarddb.com/perplexity

### Photoroom
- Website: https://www.photoroom.com
- Description: Photoroom — AI Creative
- StandardDB page: https://standarddb.com/photoroom

### Pika
- Website: https://pika.art
- Description: Pika — AI Creative
- StandardDB page: https://standarddb.com/pika

### Pinecone
- Website: https://docs.pinecone.io
- Description: Pinecone — AI Infrastructure
- StandardDB page: https://standarddb.com/pinecone

### PostHog
- Website: https://posthog.com
- Description: PostHog is an open-source product analytics platform. Features include analytics, session replay, feature flags, A/B testing, and surveys.
- StandardDB page: https://standarddb.com/posthog

### Qdrant
- Website: https://github.com
- Description: Qdrant — AI Infrastructure
- StandardDB page: https://standarddb.com/qdrant

### Railway
- Website: https://railway.com
- Description: Railway — Developer Platform
- StandardDB page: https://standarddb.com/railway

### Ramp
- Website: https://ramp.com
- Description: Ramp — Fintech / Legal
- StandardDB page: https://standarddb.com/ramp

### Raycast
- Website: https://www.raycast.com
- Description: Raycast — Developer Platform
- StandardDB page: https://standarddb.com/raycast

### Reka AI
- Website: https://www.reka.ai
- Description: Reka AI — Frontier AI Lab
- StandardDB page: https://standarddb.com/reka

### Replicate
- Website: https://replicate.com
- Description: Replicate — AI Infrastructure
- StandardDB page: https://standarddb.com/replicate

### Replit
- Website: https://blog.replit.com
- Description: Replit — AI Code Editor
- StandardDB page: https://standarddb.com/replit

### Runway
- Website: https://runwayml.com
- Description: Runway — AI Creative
- StandardDB page: https://standarddb.com/runway

### Sentry
- Website: https://sentry.io
- Description: Sentry — Developer Platform
- StandardDB page: https://standarddb.com/sentry

### Sim
- Website: https://www.sim.ai
- Description: Sim is an open-source platform for building and deploying AI agent workflows with a visual drag-and-drop canvas.
- StandardDB page: https://standarddb.com/sim

### Slack
- Website: https://docs.slack.dev
- Description: Slack — Developer Platform
- StandardDB page: https://standarddb.com/slack

### Sourcegraph
- Website: https://sourcegraph.com
- Description: Sourcegraph — AI Code Editor
- StandardDB page: https://standarddb.com/sourcegraph

### Stability AI
- Website: https://stability.ai
- Description: Stability AI — Frontier AI Lab
- StandardDB page: https://standarddb.com/stability

### StackBlitz (Bolt.new)
- Website: https://stackblitz.com
- Description: StackBlitz (Bolt.new) — AI App Builder
- StandardDB page: https://standarddb.com/bolt

### Stripe
- Website: https://stripe.com
- Description: Stripe is a financial infrastructure platform for businesses. Millions of companies use Stripe to accept payments, grow their revenue, and accelerate new business opportunities.
- StandardDB page: https://standarddb.com/stripe

### Suno
- Website: https://suno.com
- Description: Suno — AI Creative
- StandardDB page: https://standarddb.com/suno

### Supabase
- Website: https://supabase.com
- Description: Supabase is an open-source Firebase alternative. It provides a Postgres database, authentication, instant APIs, edge functions, realtime subscriptions, and storage.
- StandardDB page: https://standarddb.com/supabase

### Superhuman
- Website: https://blog.superhuman.com
- Description: Superhuman — Productivity
- StandardDB page: https://standarddb.com/superhuman

### Synthesia
- Website: https://www.synthesia.io
- Description: Synthesia — AI Creative
- StandardDB page: https://standarddb.com/synthesia

### Tabnine
- Website: https://docs.tabnine.com
- Description: Tabnine — AI Code Editor
- StandardDB page: https://standarddb.com/tabnine

### Together AI
- Website: https://docs.together.ai
- Description: Together AI — AI Infrastructure
- StandardDB page: https://standarddb.com/together

### Trigger.dev
- Website: https://trigger.dev
- Description: Trigger.dev is an open-source platform for building and deploying background jobs, AI agent workflows, and durable serverless functions in TypeScript.
- StandardDB page: https://standarddb.com/trigger-dev

### Unsloth
- Website: https://unsloth.ai
- Description: Unsloth — AI Infrastructure
- StandardDB page: https://standarddb.com/unsloth

### Val Town
- Website: https://blog.val.town
- Description: Val Town — Developer Platform
- StandardDB page: https://standarddb.com/val-town

### Vercel
- Website: https://vercel.com
- Description: Vercel is the platform for frontend developers, providing the speed and reliability innovators need to create at the moment of inspiration.
- StandardDB page: https://standarddb.com/vercel

### vLLM
- Website: https://github.com
- Description: vLLM — AI Infrastructure
- StandardDB page: https://standarddb.com/vllm

### Warp
- Website: https://docs.warp.dev
- Description: Warp — AI Code Editor
- StandardDB page: https://standarddb.com/warp

### Weaviate
- Website: https://weaviate.io
- Description: Weaviate — AI Infrastructure
- StandardDB page: https://standarddb.com/weaviate

### Weights & Biases
- Website: https://docs.wandb.ai
- Description: Weights & Biases — AI Infrastructure
- StandardDB page: https://standarddb.com/wandb

### Windsurf
- Website: https://windsurf.com
- Description: Windsurf — AI Code Editor
- StandardDB page: https://standarddb.com/windsurf

### xAI
- Website: https://x.ai
- Description: Elon Musk's AI company, building Grok
- StandardDB page: https://standarddb.com/xai

### Zapier
- Website: https://zapier.com
- Description: Zapier — Agents & Automation
- StandardDB page: https://standarddb.com/zapier

### Zed
- Website: https://zed.dev
- Description: Zed — AI Code Editor
- StandardDB page: https://standarddb.com/zed

## Products

### Anthropic (Anthropic)
- Website: 
- Description: Anthropic general updates
- StandardDB page: https://standarddb.com/product/anthropic

### Artie (Artie)
- Website: 
- Description: Artie general updates
- StandardDB page: https://standarddb.com/product/artie

### Artie Transfer (Artie)
- Website: https://www.artie.com
- Description: A database replication tool leveraging CDC to stream production data from PostgreSQL, MySQL, and MongoDB to Snowflake, BigQuery, Redshift, and Databricks in real-time.
- StandardDB page: https://standarddb.com/product/artie-transfer

### ChatGPT (OpenAI)
- Website: https://chat.openai.com
- Description: ChatGPT is a conversational AI assistant built by OpenAI, powered by the GPT family of models.
- StandardDB page: https://standarddb.com/product/chatgpt

### Claude (Anthropic)
- Website: https://claude.ai
- Description: Claude is Anthropic's frontier AI assistant, designed to be helpful, harmless, and honest.
- StandardDB page: https://standarddb.com/product/claude

### Claude Code (Anthropic)
- Website: https://claude.ai/code
- Description: Claude Code is Anthropic's agentic coding tool that lives in the terminal.
- StandardDB page: https://standarddb.com/product/claude-code

### Cloudflare (Cloudflare)
- Website: 
- Description: Cloudflare general updates
- StandardDB page: https://standarddb.com/product/cloudflare

### Cloudflare Workers (Cloudflare)
- Website: https://workers.cloudflare.com
- Description: Cloudflare Workers provides a serverless execution environment that allows you to create new applications or augment existing ones without configuring or maintaining infrastructure.
- StandardDB page: https://standarddb.com/product/cloudflare-workers

### Concourse (Concourse)
- Website: https://www.concourse.co
- Description: An AI agent platform for corporate finance that lets teams chat with financial data, automate closings, generate reports, and run scenario projections.
- StandardDB page: https://standarddb.com/product/concourse

### Gemini (Google)
- Website: https://gemini.google.com
- Description: Gemini is Google's most capable AI model family, built from the ground up to be multimodal.
- StandardDB page: https://standarddb.com/product/gemini

### GitHub (GitHub)
- Website: 
- Description: GitHub general updates
- StandardDB page: https://standarddb.com/product/github

### GitHub Actions (GitHub)
- Website: https://github.com/features/actions
- Description: Automate your workflow from idea to production. GitHub Actions makes it easy to automate all your software workflows with world-class CI/CD.
- StandardDB page: https://standarddb.com/product/github-actions

### GitHub Copilot (GitHub)
- Website: https://github.com/features/copilot
- Description: GitHub Copilot is an AI pair programmer that helps you write code faster with less work.
- StandardDB page: https://standarddb.com/product/github-copilot

### GitHub Dependabot (GitHub)
- Website: 
- Description: Automated dependency updates
- StandardDB page: https://standarddb.com/product/github-dependabot

### Google (Google)
- Website: 
- Description: Google general updates
- StandardDB page: https://standarddb.com/product/google

### Grok API (xAI)
- Website: https://docs.x.ai
- Description: xAI's Grok large language model API
- StandardDB page: https://standarddb.com/product/grok-api

### Linear (Linear)
- Website: https://linear.app
- Description: Linear helps streamline software projects, sprints, tasks, and bug tracking. Built for modern product teams.
- StandardDB page: https://standarddb.com/product/linear

### Magic Patterns (Magic Patterns)
- Website: https://www.magicpatterns.com
- Description: An AI prototyping platform for faster UI generation and iteration. Generates UI components from prompts with Figma and design system integration.
- StandardDB page: https://standarddb.com/product/magic-patterns

### Next.js (Vercel)
- Website: https://nextjs.org
- Description: The React Framework for the Web. Used by some of the world's largest companies, Next.js enables you to create high-quality web applications with the power of React components.
- StandardDB page: https://standarddb.com/product/nextjs

### OpenAI (OpenAI)
- Website: 
- Description: OpenAI general updates
- StandardDB page: https://standarddb.com/product/openai

### OpenAI API (OpenAI)
- Website: https://platform.openai.com
- Description: The OpenAI API provides access to GPT and other models for text generation, embeddings, image generation, and more.
- StandardDB page: https://standarddb.com/product/openai-api

### PostHog (PostHog)
- Website: https://posthog.com
- Description: PostHog is the all-in-one platform for product analytics, feature flags, session replays, A/B testing, and surveys.
- StandardDB page: https://standarddb.com/product/posthog

### Sim (Sim)
- Website: 
- Description: Sim general updates
- StandardDB page: https://standarddb.com/product/sim

### Sim Studio (Sim)
- Website: https://www.sim.ai
- Description: Sim Studio is an open-source visual AI agent workflow builder. Design agent workflows on a canvas, connect tools and blocks, then run them instantly.
- StandardDB page: https://standarddb.com/product/sim-studio

### Stripe (Stripe)
- Website: 
- Description: Stripe general updates
- StandardDB page: https://standarddb.com/product/stripe

### Stripe API (Stripe)
- Website: https://stripe.com/docs/api
- Description: The Stripe API is organized around REST. It has predictable resource-oriented URLs, accepts form-encoded request bodies, and returns JSON-encoded responses.
- StandardDB page: https://standarddb.com/product/stripe-api

### Stripe Billing (Stripe)
- Website: https://stripe.com/billing
- Description: Stripe Billing lets you bill and manage customers however you want — from simple recurring billing to usage-based billing and sales-negotiated contracts.
- StandardDB page: https://standarddb.com/product/stripe-billing

### Supabase (Supabase)
- Website: https://supabase.com
- Description: Supabase provides a Postgres database, authentication, instant APIs, edge functions, realtime subscriptions, storage, and vector embeddings.
- StandardDB page: https://standarddb.com/product/supabase

### Trigger.dev (Trigger.dev)
- Website: https://trigger.dev
- Description: An open-source background jobs and task orchestration platform for TypeScript with no timeouts, retries, queues, and elastic scaling.
- StandardDB page: https://standarddb.com/product/trigger-dev

### Vercel (Vercel)
- Website: 
- Description: Vercel general updates
- StandardDB page: https://standarddb.com/product/vercel

### Vercel AI SDK (Vercel)
- Website: https://sdk.vercel.ai
- Description: An open-source library for building AI-powered user interfaces. The Vercel AI SDK provides React hooks and streaming utilities for LLM applications.
- StandardDB page: https://standarddb.com/product/vercel-ai-sdk

## Changelog Entries

### v0.5.103
- Published: 2026-03-02T07:37:46+00:00
- Products: Sim Studio
- Tags: release, feature, api, integration
- URL: https://standarddb.com/sim/sim-studio-v0-5-103-adds-openapi-docs-and-three-new-integrations--qjpxlmgq
- Original: https://github.com/simstudioai/sim/releases/tag/v0.5.103

## API Documentation Improvements

Sim Studio v0.5.103 introduces comprehensive API reference documentation powered by an OpenAPI specification. This new documentation includes auto-generated endpoint pages, making it easier for developers to understand and integrate with Sim Studio's APIs.

## New Integrations

The release expands Sim Studio's integration ecosystem with three new connectors:

- **Amplitude**: Native integration for event analytics and behavioral data
- **Google PageSpeed Insights**: Monitor and analyze web performance metrics
- **PagerDuty**: Enable incident management and alert orchestration workflows

## Bug Fixes

A production monitoring fix updates the MemoryTelemetry logger to INFO level, improving visibility into memory usage patterns in production environments.

## Action Items

Developers using Sim Studio should review the new API documentation to understand available endpoints. Teams implementing monitoring or analytics workflows can now leverage the new Amplitude, PageSpeed Insights, and PagerDuty integrations directly.

---

### New Plans, Credits, and On-Demand Pricing
- Published: 2026-03-01T00:00:00+00:00
- Products: Magic Patterns
- Tags: pricing, feature, platform
- URL: https://standarddb.com/magic-patterns/magic-patterns-launches-usage-based-pricing-with-on-demand-billing-and-feature--gyzckpdg
- Original: https://www.magicpatterns.com/blog/new-plans-and-pricing

## Usage-Based Credit System

Magic Patterns is replacing its "one prompt, one credit" model with a usage-based credit system that scales based on request complexity. Credits now map to underlying AI inference costs (input and output tokens), aligning with industry standards like Cursor. In testing, 1 credit in the old system maps to approximately 10 credits in the new system. Free actions like visual edits, "Fix with AI," and retries remain unmetered.

## On-Demand Pricing and Budget Controls

All paid plans will support optional on-demand pricing starting April 2, allowing usage beyond monthly limits to be billed at a fixed per-credit rate without requiring plan upgrades. Users can set budget limits and alert thresholds to maintain control over spending.

## New Feature-Based Plan Tiers

Plans are now organized by capabilities rather than credit limits:

- **Starter** (formerly Hobby): Design systems, shared workspace, centralized billing, on-demand pricing
- **Business** (formerly Pro): Starter features plus faster AI models, self-serve SSO, user roles & permissions, usage reporting, team workspaces (11+ users)
- **Enterprise**: All features plus technical account manager, priority support, shared credit pooling, SCIM & audit logs, custom contracts

## Migration and Pricing Timeline

Existing customers migrate automatically on April 2 with pricing preserved:
- Monthly subscribers ($19/$75) keep current rates until June 30, 2026, then move to standard pricing ($20/$100)
- Annual subscribers maintain current pricing through their cycle end, then renew at new rates
- All customers can upgrade, downgrade, or cancel anytime

No action is required from existing subscribers.

---

### v0.17.5
- Published: 2026-02-28T22:21:42+00:00
- Products: 
- Tags: release, feature, open-source
- URL: https://standarddb.com/ollama/ollama-v0-17-5-adds-gguf-model-compatibility-support--bpfjbzpm
- Original: https://github.com/ollama/ollama/releases/tag/v0.17.5

## Ollama v0.17.5 Released

Ollama has released version 0.17.5, focusing on expanding model compatibility and improving the user experience for developers working with quantized models.

### Key Changes

**GGUF Model Compatibility** - This release adds compatibility with imported GGUF models, allowing users to leverage a broader ecosystem of pre-quantized models. GGUF is a popular format for optimized model inference, and this enhancement makes it easier to integrate external models into Ollama's workflow.

### What Developers Need to Know

- You can now import and run GGUF format models more seamlessly within Ollama
- This expands the available model ecosystem without requiring format conversion
- The change maintains compatibility with existing workflows and models

### How to Get Started

Download Ollama v0.17.5 from the official repository to access the improved GGUF compatibility features.

---

### v0.5.102
- Published: 2026-02-28T20:56:37+00:00
- Products: Sim Studio
- Tags: release, integration, feature, performance, bugfix
- URL: https://standarddb.com/sim/sim-v0-5-102-adds-10-new-integrations-including-x-api-v2-databricks-and--hhxzqnpt
- Original: https://github.com/simstudioai/sim/releases/tag/v0.5.102

## New Integrations

Sim v0.5.102 expands the platform's capability with 10 new integrations across communication, recruiting, analytics, and event management:

- **X (Twitter) API v2**: Added 28 new tools with expanded OAuth scopes for more granular access control
- **Recruiting platforms**: Ashby and Greenhouse integrations for managing candidates, jobs, and applications
- **Email & communications**: Loops email platform integration; Resend expanded with contacts, domains, and enhanced email operations
- **Event management**: Luma integration for event and guest management
- **Analytics & data**: Databricks integration with 8 tools for data operations
- **Content generation**: Gamma AI-powered content generation integration
- **CRM**: Google Contacts integration

## Agent Improvements

The release introduces **MCP server discovery mode** for agents, allowing tool input selection from discovered MCP servers. The platform now adds individual MCP server tools rather than single server entries, providing more granular control over agent capabilities.

## Performance & Stability

- Fixed memory leaks in SSE stream cleanup with added memory telemetry
- Improved unit test performance by eliminating `vi.resetModules` anti-pattern
- Enhanced CI/CD pipeline with sticky disk caches and faster build runners
- Fixed chat deployment auth persistence issues and cleaned up React anti-patterns

## Developer Experience

- Consolidated selector input logic for cleaner implementation patterns
- Made selector keys declarative for easier maintenance
- Reordered OAuth modal for better user flow
- Standardized tool descriptions across documentation, with developer count now at 70k

---

### Copilot usage metrics now includes enterprise-level GitHub Copilot CLI activity
- Published: 2026-02-27T18:00:19+00:00
- Products: GitHub Copilot, GitHub Dependabot
- Tags: api, feature, release
- URL: https://standarddb.com/github/github-copilot-usage-metrics-now-track-cli-activity-for-enterprise-deployments--wjvfrqhm
- Original: https://github.blog/changelog/2026-02-27-copilot-usage-metrics-now-includes-enterprise-level-github-copilot-cli-activity

## CLI Metrics Now Available in Enterprise Dashboard

GitHub's Copilot usage metrics API now includes comprehensive telemetry from the Copilot CLI, extending enterprise visibility beyond web and IDE integrations. Organizations can now access CLI-specific metrics through the same REST API endpoint used for other Copilot telemetry.

## New Metrics Available

The expanded metrics include:

- **Daily active CLI users** — Track adoption and engagement patterns among developers using the CLI
- **CLI request and session counts** — Understand usage frequency and user behavior in terminal environments
- **Token usage totals** — Monitor consumption including average tokens per request for capacity planning

## Use Cases for Teams

Organizations can leverage these new metrics to:

- **Identify adoption gaps** — Compare CLI adoption across teams and departments to determine where enablement or additional support is needed
- **Understand developer workflows** — Compare usage patterns between CLI, IDE, and web interfaces to see how different teams prefer to work
- **Plan consumption and rollout** — Use token usage data to forecast costs and phase deployments strategically across the enterprise

This update builds on GitHub's recent push to provide granular, actionable metrics for enterprise Copilot deployments, complementing recent releases around metrics availability and enterprise AI controls.

## Action Items

Enterprises with GitHub Copilot can begin querying the metrics API to retrieve CLI activity data through the existing [Copilot usage metrics endpoint](https://docs.github.com/enterprise-cloud@latest/rest/copilot/copilot-usage-metrics). No additional setup is required for data collection, as CLI telemetry is automatically captured.

---

### Develop Native Multimodal Agents with Qwen3.5 VLM Using NVIDIA GPU-Accelerated Endpoints
- Published: 2026-02-27T17:30:00+00:00
- Products: 
- Tags: feature, model, api, platform, integration
- URL: https://standarddb.com/nvidia/nvidia-launches-free-gpu-accelerated-endpoints-for-alibaba-s-qwen3-5-397b--kmgkmcgq
- Original: https://developer.nvidia.com/blog/develop-native-multimodal-agents-with-qwen3-5-vlm-using-nvidia-gpu-accelerated-endpoints

## Qwen3.5 Now Available on NVIDIA Infrastructure

Alibaba has released Qwen3.5, a 397-billion parameter native vision-language model designed for multimodal agents. The model uses a hybrid architecture combining mixture of experts (MoE) with Gated Delta Networks, achieving an impressive 4.28% activation rate with only 17B active parameters per token. It supports 256K token context windows (extensible to 1M), covers 200+ languages, and can understand and navigate complex user interfaces.

## Immediate Access for Developers

Developers can start building with Qwen3.5 immediately through multiple channels:

- **Free GPU-accelerated endpoints** on build.nvidia.com powered by NVIDIA Blackwell GPUs, available to registered NVIDIA Developer Program members
- **API access** through NVIDIA's hosted endpoints with free usage tier
- **Full code examples** and OpenAI-compatible chat completion APIs for rapid integration

The model excels at coding tasks, visual reasoning over mobile and web interfaces, chat applications, and complex search scenarios.

## Production Deployment and Customization

For production use, **NVIDIA NIM** provides containerized inference microservices with optimized performance, standardized APIs, and deployment flexibility across on-premises, cloud, and hybrid environments.

The **NVIDIA NeMo framework** enables fine-tuning for specialized domains. Key capabilities include:

- PyTorch-native training with Day 0 Hugging Face checkpoint support (no conversion needed)
- Memory-efficient methods like LoRA for cost-effective adaptation
- Multinode deployment on Slurm and Kubernetes for large-scale training
- Reference implementation for medical visual QA and radiological dataset fine-tuning

## Getting Started

Developers can access Qwen3.5 immediately on build.nvidia.com, experiment with prompts, and test against their own data. Integration with existing NVIDIA infrastructure (Blackwell GPUs, NIM, NeMo) enables seamless scaling from prototyping to enterprise production workloads.

---

### Copilot metrics is now generally available
- Published: 2026-02-27T17:29:13+00:00
- Products: GitHub Copilot, GitHub Dependabot
- Tags: release, feature, api, platform
- URL: https://standarddb.com/github/github-copilot-metrics-now-generally-available-with-enterprise-dashboards-and--gqswzqph
- Original: https://github.blog/changelog/2026-02-27-copilot-metrics-is-now-generally-available

## Overview

GitHub Copilot usage metrics has graduated from public preview to general availability, providing organizations with a comprehensive platform for understanding how teams adopt and use Copilot. The feature enables IT leaders and decision-makers to track trends, assess enablement efforts, and build custom reports tailored to their organizational structure.

## What's Included

**Dashboards and Visibility**

The GA release includes three new dashboards that provide multi-level insights:

- **Copilot usage dashboard**: Tracks code completion activity, IDE usage, model and language breakdown, and overall team adoption patterns
- **Code generation dashboard**: Quantifies Copilot's output by measuring suggested, added, and deleted lines across completions, chat, and agent features
- **Multi-level granularity**: Enterprise and organization owners can now view adoption trends without relying solely on enterprise-wide aggregates

**API Capabilities**

Three-tier API access enables programmatic access to usage data:

- **Enterprise-level**: Aggregate analysis across all Copilot users for bird's-eye view of engagement and pull request throughput
- **Organization-level**: Teams can programmatically access their own Copilot usage metrics to spot trends and assess enablement
- **User-level**: Daily individual usage analysis to identify teams needing training or documentation support

## Access and Controls

Access is available through **Enterprise Account → AI Controls → Copilot → Metrics**. Fine-grained access controls allow custom enterprise roles (such as `View enterprise Copilot metrics`) without requiring full admin access. GitHub Enterprise Cloud customers with data residency requirements can now access both dashboards and API reporting.

## What's Next

GitHub is building toward impact measurement, connecting usage patterns to engineering outcomes to help organizations move beyond adoption tracking to demonstrating Copilot's business value.

---

### Vercel Queues now in public beta
- Published: 2026-02-27T13:00:00+00:00
- Products: Vercel
- Tags: feature, release, api, platform
- URL: https://standarddb.com/vercel/vercel-queues-enters-public-beta-with-durable-event-streaming-and-at-least-once--txcwwqdq
- Original: https://vercel.com/changelog/vercel-queues-now-in-public-beta

## Vercel Queues Public Beta Launch

Vercel has moved Vercel Queues out of experimental status and into public beta, making it available to all teams. This durable event streaming system is built atop Vercel's Fluid compute infrastructure and addresses a critical need: reliably handling asynchronous work and guaranteeing that long-running tasks complete successfully.

### How It Works

Queues follows a clean pub/sub model:

- **Message publishing**: Send messages from any route handler to a durable topic
- **Fan-out delivery**: The queue distributes messages to subscribed consumer groups
- **Independent processing**: Each consumer group processes messages independently with automatic retries
- **Guaranteed delivery**: Messages are redelivered until successfully processed or expired

The system provides at-least-once delivery semantics with built-in handling for common failure scenarios like function crashes and new deployments rolling out.

### Core Features

Vercel Queues includes:

- Multiple availability zone synchronous replication
- Customizable visibility timeout and delayed delivery
- Idempotency keys for deduplication
- Concurrency control per consumer
- Per-deployment topic partitioning

Consumer routes configured via `vercel.json` become private endpoints—they have no public URL and can only be invoked by Vercel's queue infrastructure.

### Pricing and Integration

Queues billing starts at **$0.60 per 1M API operations**, making it accessible for a range of workloads. Functions invoked by Queues in push mode use existing Fluid compute rates. Queues also powers Vercel's Workflow feature, which provides higher-level multi-step orchestration on top of the messaging primitives.

Developers can get started with the [Queues documentation](https://vercel.com/docs/queues) and deploy using a free Vercel account.

---

### ASPA: making Internet routing more secure
- Published: 2026-02-27T06:00:00+00:00
- Products: Cloudflare
- Tags: feature, security, platform
- URL: https://standarddb.com/cloudflare/cloudflare-radar-adds-aspa-monitoring-to-detect-bgp-route-leaks-in-real-time--hsfrzfpd
- Original: https://blog.cloudflare.com/aspa-secure-internet

## What is ASPA?

ASPA (Autonomous System Provider Authorization) is a new cryptographic standard designed to secure the path Internet traffic takes across networks. While existing RPKI (Resource Public Key Infrastructure) technology verifies that traffic arrives at the correct destination through Route Origin Authorizations (ROAs), ASPA validates the entire journey by allowing networks to publish a list of their authorized upstream providers.

When data crosses the Internet, it maintains a running log of every network it passes through—called the `AS_PATH` in BGP (Border Gateway Protocol). ASPA enables receiving networks to cryptographically verify that traffic only traveled through an approved chain of providers, preventing **route leaks** where traffic gets misdirected through unintended networks.

## How ASPA Detection Works

ASPA relies on the hierarchical structure of Internet routing, often visualized as a "valley-free" model:
- **Up-Ramp:** Traffic travels from a customer "up" through increasingly larger providers
- **Apex:** Traffic may cross a single peering link between major providers
- **Down-Ramp:** Traffic flows "down" through providers to the destination

Route leaks typically manifest as "valleys"—when traffic unexpectedly goes down to a customer and then back up to another provider. ASPA validation checks both the up and down paths to ensure they meet at the apex with continuous authorization. If the paths don't connect, ASPA flags the route as invalid.

## Cloudflare Radar Integration

Cloudflare has integrated ASPA monitoring into Cloudflare Radar, providing visibility into:
- ASPA adoption trends across the five Regional Internet Registries (RIRs)
- ASPA records at the Autonomous System (AS) level
- Historical changes to ASPA deployments over time

This tooling helps the network community track ASPA rollout and adoption as the industry transitions to more secure routing practices.

---

### Bringing more transparency to post-quantum usage, encrypted messaging, and routing security
- Published: 2026-02-27T06:00:00+00:00
- Products: Cloudflare
- Tags: feature, security, api, platform
- URL: https://standarddb.com/cloudflare/cloudflare-radar-adds-post-quantum-monitoring-for-origin-connections-key--sthqxqch
- Original: https://blog.cloudflare.com/radar-origin-pq-key-transparency-aspa

## Post-Quantum Encryption Monitoring Extends to Origin Servers

Cloudflare Radar previously tracked post-quantum (PQ) adoption only for client-to-Cloudflare connections, which has grown from under 3% in early 2024 to over 60% in February 2026. Today's update extends this monitoring to **origin-facing connections**—the critical link between Cloudflare's edge servers and customer origin servers.

The new origin post-quantum support graph visualizes the percentage of customer origins supporting the `X25519MLKEM768` hybrid key exchange algorithm, which combines classical X25519 encryption with ML-KEM, a NIST-standardized lattice-based post-quantum scheme. Data is sourced from Cloudflare's automated TLS scanner, which probes TLS 1.3-compatible origins daily. Currently, approximately **10% of origins support post-quantum encryption**—a dramatic 10x increase from less than 1% at the start of 2025. The full Data Explorer view on Radar shows support across all classical and quantum-resistant key exchange methods.

## Key Transparency Dashboard for Encrypted Messaging

A new **Key Transparency section** on Radar provides a public, real-time dashboard tracking the verification status of Key Transparency Logs for end-to-end encrypted messaging services like WhatsApp. The dashboard displays:

- When each log was last signed and verified by Cloudflare's Auditor
- Real-time monitoring of the integrity of public key distribution
- Public API access for independent validation of Auditor proofs

This transparency tool enables anyone to audit the security of messaging service key distribution infrastructure.

## Expanded Routing Security Metrics

Radar's Routing Security insights now include global, country, and network-level deployment data for **ASPA** (Autonomous System Provider Authorization), an emerging BGP standard that helps detect and prevent route leaks. This addition strengthens visibility into Internet routing security posture across regions and autonomous systems.

## What Developers Need to Know

Organizations should monitor their origin post-quantum readiness on Radar and work with infrastructure teams to enable support where applicable, especially as the ecosystem continues its migration toward quantum-resistant cryptography. The scanner tests for *support* rather than preference, so origins may support PQ algorithms without preferring them by default.

---

### OpenAI and Amazon announce strategic partnership
- Published: 2026-02-27T05:30:00+00:00
- Products: OpenAI API, OpenAI
- Tags: release, feature, integration, platform, partnership
- URL: https://standarddb.com/openai/openai-and-amazon-launch-stateful-runtime-environment-on-aws-bedrock-50b--clsljsrn
- Original: https://openai.com/index/amazon-partnership

## New Products and Services

OpenAI and Amazon are jointly developing a **Stateful Runtime Environment** powered by OpenAI's models, available through Amazon Bedrock. This enables developers to build AI applications and agents that can keep context, access compute resources, maintain memory across sessions, and work seamlessly with software tools and data sources. The environment is optimized to run on AWS infrastructure and integrates with Amazon Bedrock's AgentCore, launching in the coming months.

AWS will serve as the **exclusive third-party cloud distribution provider** for OpenAI Frontier, OpenAI's enterprise platform for building, deploying, and managing teams of AI agents at production scale. Frontier enables organizations to build AI agents that operate across business systems with shared context, governance, and enterprise-grade security—eliminating the need to manage underlying infrastructure.

## Infrastructure and Investment

OpenAI and AWS are expanding their existing $38 billion multi-year agreement by an additional $100 billion over 8 years. OpenAI will consume approximately **2 gigawatts of Trainium capacity** through AWS infrastructure to support the Stateful Runtime Environment, Frontier, and advanced workloads. This commitment spans both Trainium3 and next-generation Trainium4 chips (launching 2027), providing long-term compute capacity while reducing costs and improving efficiency.

Amazon will invest **$50 billion in OpenAI**, starting with an initial $15 billion investment followed by $35 billion in subsequent months pending certain conditions being met.

## Custom Models and Developer Access

OpenAI and Amazon will collaborate to develop **customized models** available to Amazon developers for powering Amazon's customer-facing applications and agents. These offerings complement Amazon's existing Nova family of models, providing Amazon teams with additional tools for building and scaling AI products.

Developers using AWS will gain access to these new capabilities alongside existing OpenAI models through Amazon Bedrock, simplifying enterprise AI deployment and management.

---

### Introducing the Stateful Runtime Environment for Agents in Amazon Bedrock
- Published: 2026-02-27T05:30:00+00:00
- Products: OpenAI API, OpenAI
- Tags: feature, api, platform, integration, release
- URL: https://standarddb.com/openai/openai-and-amazon-launch-stateful-runtime-environment-for-production-grade--pfnqydrw
- Original: https://openai.com/index/introducing-the-stateful-runtime-environment-for-agents-in-amazon-bedrock

## Stateful Runtime for Production Agents

OpenAI and Amazon today announced the **Stateful Runtime Environment**, a jointly developed solution that brings production-grade agent capabilities to Amazon Bedrock. The runtime is powered by OpenAI models and optimized for AWS infrastructure, addressing a critical gap in agentic workflows: the operational complexity of running reliable, multi-step work with proper state management and governance.

## What Changed

The new runtime operates **natively inside your AWS environment** and handles the orchestration challenges that have historically required custom development:

- **Persistent state management** across multi-step workflows
- **Working context** that carries forward memory, history, and tool outputs
- **Automatic error handling** and safe resumption of long-running tasks
- **Identity and permission boundaries** for secure agent execution
- **AWS-native governance** integrated with existing security posture and tooling

## Why It Matters

Previous agent deployments relied on stateless APIs, forcing development teams to build their own orchestration layers. Production workflows require handling approval chains, multi-system interactions, context persistence, and reliable recovery—all of which required substantial scaffolding. This runtime eliminates that burden by providing a purpose-built execution environment.

Developers can now focus on business logic and workflow design rather than infrastructure, significantly reducing time to production for use cases like customer support automation, sales operations, IT automation, and finance processes with audits.

## Getting Started

The Stateful Runtime Environment in Amazon Bedrock will be available soon. AWS customers can request early access through their OpenAI account teams.

---

### v0.17.3
- Published: 2026-02-27T04:04:56+00:00
- Products: 
- Tags: release, bugfix
- URL: https://standarddb.com/ollama/ollama-v0-17-3-fixes-tool-calling-bug-in-qwen-3-and-3-5-models--wfdhlfwp
- Original: https://github.com/ollama/ollama/releases/tag/v0.17.3

## Bug Fix

Ollama v0.17.3 resolves a parsing issue that affected tool calling capabilities in Qwen 3 and Qwen 3.5 model families. The bug specifically occurred when tool calls were emitted during the model's extended thinking process, causing the calls to be parsed incorrectly and preventing proper execution.

## What You Need to Know

If you're using Ollama with Qwen 3 or Qwen 3.5 models and relying on tool/function calling features, this update is recommended. The fix ensures that tool calls are correctly parsed regardless of whether they occur during normal inference or during the model's thinking phase.

## Action Items

- **Update recommended**: Pull the latest v0.17.3 release to restore proper tool calling functionality for Qwen models
- **No breaking changes**: This is a backward-compatible patch with no API changes required

---

### v0.17.2
- Published: 2026-02-27T02:24:54+00:00
- Products: 
- Tags: release, bugfix
- URL: https://standarddb.com/ollama/ollama-v0-17-2-fixes-windows-app-crash-on-update-downloads--zxbpgrhz
- Original: https://github.com/ollama/ollama/releases/tag/v0.17.2

## Ollama v0.17.2 Release

**What's Fixed**

This patch release addresses a stability issue in Ollama's Windows application that caused crashes when the app attempted to download available updates. Windows users who experienced unexpected crashes during update checks should update to this version immediately.

**Who This Affects**

Windows users running Ollama, particularly those with automatic update checking enabled, were affected by this issue. The crash prevented users from completing the update process and could interrupt regular usage.

**Action Items**

- **Windows users**: Update to v0.17.2 to resolve the crash issue
- Download the latest release from the [GitHub releases page](https://github.com/ollama/ollama/releases/tag/v0.17.2)

---

### February 26, 2026
- Published: 2026-02-27T00:15:10+00:00
- Products: 
- Tags: api, feature, sdk, release
- URL: https://standarddb.com/notion/notion-api-adds-markdown-content-endpoints-sdk-gains-auto-retry-and-mcp--mlnmlhpz
- Original: https://developers.notion.com/page/changelog

## Markdown Content API

Notion introduced three new API endpoints that let developers work with page content using enhanced markdown format instead of the traditional block-based approach:

- **POST /v1/pages** now accepts a `markdown` parameter as an alternative to the `children` parameter for creating pages with markdown content
- **GET /v1/pages/:page_id/markdown** retrieves a page's full content formatted as enhanced markdown
- **PATCH /v1/pages/:page_id/markdown** allows inserting or replacing content using markdown with ellipsis-based selections for granular updates

The SDK adds corresponding helper methods `pages.retrieveMarkdown()` and `pages.updateMarkdown()` for easier integration. Complete documentation is available in the [Working with markdown content](https://developers.notion.com/guides/data-apis/working-with-markdown-content) guide.

## SDK Improvements

The JavaScript/TypeScript SDK now includes **automatic retry with exponential backoff** for failed requests, configurable to match your application's needs. This improves reliability for integrations handling transient failures.

## AI Meeting Notes & MCP Enhancements

The markdown retrieval endpoint supports an `include_transcript` query parameter to include full meeting note transcripts. Notion MCP received significant improvements including ~91% context token reduction in database and data source tools by switching to SQL DDL-based schemas, new tools for querying meeting notes and fetching data sources by ID, and enterprise governance features including audit logging and admin tool allowlisting.

## Action Items

- Review the [Enhanced markdown format reference](https://developers.notion.com/guides/data-apis/enhanced-markdown) to understand the new markdown syntax
- Update your SDK to v5.10.0+ to benefit from automatic retry functionality
- If using Notion MCP with third-party AI tools, reconnect to ensure you have the latest tools and reduced token consumption
- Consult the [security best practices guide](https://developers.notion.com/docs/mcp-security-best-practices) when deploying MCP integrations

---

### Radar - Post-Quantum Encryption and Key Transparency on Cloudflare Radar
- Published: 2026-02-27T00:00:00+00:00
- Products: Cloudflare
- Tags: release, feature, api, security, platform
- URL: https://standarddb.com/cloudflare/cloudflare-radar-adds-post-quantum-encryption-tracking-and-key-transparency--fvgsjvfp
- Original: https://developers.cloudflare.com/changelog/post/2026-02-27-radar-pq-key-transparency

## Post-Quantum Encryption Tracking

**Cloudflare Radar** now includes comprehensive tools for monitoring post-quantum cryptography adoption. A new Post-Quantum Encryption page visualizes the share of customer origins supporting X25519MLKEM768, derived from daily automated TLS scans. This data is sourced from testing TLS 1.3-compatible origins to measure actual algorithm support capabilities.

### New Post-Quantum APIs

Three new endpoints power the post-quantum tracking features:

- **`/post_quantum/tls/support`** – Tests whether a specific host supports post-quantum TLS key exchange
- **`/post_quantum/origin/summary/{dimension}`** – Returns aggregated origin post-quantum data summarized by key agreement algorithm
- **`/post_quantum/origin/timeseries_groups/{dimension}`** – Provides time-series data grouped by key agreement algorithm

### Post-Quantum Host Test Tool

A public-facing diagnostic tool allows developers and security teams to test any publicly accessible website for post-quantum encryption compatibility. Users can enter a hostname and optional port to verify whether the server negotiates a post-quantum key exchange algorithm.

## Key Transparency Audit Dashboard

The new Key Transparency section on Radar provides visibility into audit logs for end-to-end encrypted messaging services. The dashboard currently monitors **WhatsApp** and **Facebook Messenger Transport** logs, displaying:

- Current audit status
- Last signed epoch and verified epoch timestamps
- Root hash of the Auditable Key Directory tree

This data is also accessible programmatically via the **Key Transparency Auditor API**, enabling automated monitoring and integration with security workflows.

## Getting Started

Both features are now live on [radar.cloudflare.com](https://radar.cloudflare.com). Visit the Post-Quantum Encryption and Key Transparency pages to explore the data and monitor adoption trends.

---

### Chat SDK adds Telegram adapter support
- Published: 2026-02-27T00:00:00+00:00
- Products: Vercel AI SDK, Vercel
- Tags: feature, sdk, integration
- URL: https://standarddb.com/vercel/vercel-chat-sdk-adds-telegram-adapter-now-supports-five-messaging-platforms--wmqmhsvc
- Original: https://vercel.com/changelog/chat-sdk-adds-telegram-adapter-support

## Telegram Adapter Now Available

Vercel's Chat SDK has expanded its multi-platform support to include Telegram, allowing developers to build bots that work seamlessly across five messaging platforms using a unified codebase. This addition maintains the SDK's philosophy of write-once, deploy-everywhere functionality.

## Key Features

The Telegram adapter supports:
- **Message interactions**: Mentions, reactions, and direct messages
- **Real-time feedback**: Typing indicators
- **Rich content**: Text cards with buttons and link buttons that render as inline keyboard elements
- **File handling**: Single file uploads

## Getting Started

Setting up a Telegram bot requires minimal configuration:

```typescript
import { Chat } from "chat";
import { createTelegramAdapter } from "@chat-adapter/telegram";

const bot = new Chat({
  userName: "mybot",
  adapters: {
    telegram: createTelegramAdapter(),
  },
});

bot.onNewMention(async (thread, message) => {
  await thread.post(`You said: ${message.text}`);
});
```

## Important Limitations

Developers should be aware of Telegram's platform constraints:
- **Message history**: Not exposed via full APIs; the adapter uses caching at the adapter level
- **Callback data**: Limited to 64 bytes
- **Unsupported features**: Modals and ephemeral messages are not currently available

For complete implementation details and best practices, see the [official documentation](https://www.chat-sdk.dev/docs/adapters/telegram).

---

### February 27, 2026
- Published: 2026-02-27T00:00:00+00:00
- Products: 
- Tags: feature, api, sdk
- URL: https://standarddb.com/deepgram/deepgram-flux-adds-configure-control-message-for-mid-stream-speech-recognition--mlrczxzp
- Original: https://developers.deepgram.com/changelog/2026/2/27

## Dynamic Mid-Stream Configuration Without Reconnection

Deepgram Flux now supports a **Configure control message** that enables real-time configuration updates over an active WebSocket connection. This eliminates the need to disconnect and reconnect when speech recognition behavior needs to adapt during a conversation, simplifying application logic and preventing audio loss during state transitions.

## What You Can Now Update Mid-Stream

The Configure message allows dynamic adjustments to three key parameters:

- **Keyterms** – Inject task-critical vocabulary as conversation context changes (customer names, medications, product terminology)
- **Turn detection thresholds** (`eot_threshold`, `eager_eot_threshold`) – Tighten or relax confidence requirements for detecting end of turn
- **Timeout behavior** (`eot_timeout_ms`) – Adjust silence tolerance for different conversation phases

## Real-World Use Cases

**Context-aware vocabulary:** Add a customer's name to keyterms right before requesting it, or swap to domain-specific terminology when transitioning from general discussion to pharmacy or product-specific troubleshooting.

**Adaptive turn detection:** Increase `eot_timeout_ms` and `eot_threshold` when collecting passwords, OTPs, or account numbers to prevent premature cutoffs. Relax constraints during natural conversation.

**Simplified architecture:** Manage a single persistent connection instead of handling multiple concurrent streams or reconnection state transitions.

## Implementation Details

Send a Configure control message over the Flux WebSocket connection:

```json
{
  "type": "Configure",
  "thresholds": {
    "eot_threshold": 0.8,
    "eot_timeout_ms": 5000
  },
  "keyterms": ["product_name", "feature_name"]
}
```

You'll receive a `ConfigureSuccess` response confirming the update, with changes taking effect immediately in audio stream order. **Note:** Keyterms are replaced (not merged), so include both existing and new terms if you want to add to the list.

## Availability & Next Steps

Configure is currently available via direct WebSocket connections only. SDK support and self-hosted deployment options are coming soon. See the [Configure Control Message documentation](https://developers.deepgram.com/docs/flux/configure) for comprehensive examples and the [Flux API Reference](https://developers.deepgram.com/reference/speech-to-text/listen-flux) for full specifications.

---

### v0.5.101
- Published: 2026-02-26T23:23:12+00:00
- Products: Sim Studio
- Tags: release, feature, integration, bugfix, platform
- URL: https://standarddb.com/sim/sim-releases-v0-5-101-with-devin-integration-and-expanded-google-workspace--plvsjzhv
- Original: https://github.com/simstudioai/sim/releases/tag/v0.5.101

## New Integrations & Features

Sim v0.5.101 introduces several significant integrations to expand workflow automation capabilities:

- **Devin Integration**: Enables autonomous coding sessions directly within Sim workflows, allowing users to delegate complex coding tasks to AI agents
- **Google Workspace Expansion**: Adds Google Tasks and Google Translate integrations, plus missing tools for Gmail, Drive, Sheets, and Calendar operations
- **Data Platform Integration**: Introduces Google BigQuery support for querying and managing data warehouses
- **Enhanced Confluence Tools**: New user lookup by account ID and improved page content retrieval in version tools

## Workflow Management Improvements

The release adds workflow locking capabilities with new context menu options and panel controls for locking/unlocking workflows from the sidebar. This enables better control over workflow modifications and prevents accidental changes.

## Enterprise & Admin Features

New audit log read endpoints have been added for admin and enterprise users, supporting improved compliance and monitoring capabilities.

## Bug Fixes & Optimization

Notable fixes address Confluence content preservation during updates, prevention of raw input overwriting coerced values, and improved loop support through proper execution order threading in child workflows. Database optimization includes removal of redundant indexes and addition of partial indexes for stale execution cleanup.

## Developer Considerations

The release maintains backward compatibility while expanding integration coverage. Users relying on Google Workspace tools should review newly available operations for Sheets filtering and enhanced Gmail/Drive/Calendar functionality.

---

### GitHub Actions now supports uploading and downloading non-zipped artifacts
- Published: 2026-02-26T22:36:48+00:00
- Products: GitHub Actions
- Tags: feature, platform, api
- URL: https://standarddb.com/github/github-actions-now-supports-non-zipped-artifacts-eliminating-double-compression--hnldgzxs
- Original: https://github.blog/changelog/2026-02-26-github-actions-now-supports-uploading-and-downloading-non-zipped-artifacts

## Non-Zipped Artifacts Now Available

GitHub Actions now supports uploading and downloading artifacts without automatic compression. Previously, the `actions/upload-artifact` action always zipped artifacts, requiring users to manually unzip files after download. This has been a common source of friction in CI/CD workflows.

## Key Improvements

The new functionality addresses three major pain points:

- **Single-file downloads**: Users no longer need to unzip when downloading a single file through the browser
- **Native file viewing**: Artifacts can now be viewed directly in the browser for compatible formats (HTML, images, markdown, etc.), including on mobile browsers
- **Eliminate double compression**: Pre-compressed files (like ZIP or TAR archives) no longer get wrapped in an additional compression layer

## Implementation Details

To use unzipped artifacts, you need to:

- Upgrade to `actions/upload-artifact` **v7** or later
- Set the `archive` parameter to `false` in your workflow configuration
- Update to `actions/download-artifact` **v8** or later for consistent behavior

**Backwards compatibility**: The `archive` parameter defaults to `true`, so existing workflows remain unchanged. Previously uploaded artifacts will continue to be zipped. Only new artifacts uploaded with v7 and the `archive: false` flag will be unzipped.

## Action Items

Review your GitHub Actions workflows to identify where non-zipped artifacts would improve efficiency. Update the upload and download artifact actions to v7 and v8 respectively, then set `archive: false` for artifacts that don't require compression.

---

### Copilot metrics report URLs update
- Published: 2026-02-26T20:11:34+00:00
- Products: GitHub Copilot, GitHub
- Tags: api, platform
- URL: https://standarddb.com/github/github-copilot-metrics-api-now-served-from-new-cdn-domains--nzcknwnx
- Original: https://github.blog/changelog/2026-02-26-copilot-metrics-report-urls-update

### API Migration Details

GitHub has updated the **GitHub Copilot usage metrics API** to serve report download URLs from a new CDN endpoint. The underlying data, API contract, and response schema have not changed—this is purely an infrastructure migration to improve performance and reliability.

### Action Items for Administrators

Organizations that manage firewall rules should take the following steps:

- **If you allowlist specific CDN domains**: Add the new pattern `copilot-reports-production-*.b01.azurefd.net` to your firewall allowlist alongside the existing `copilot-reports-*.b01.azurefd.net` pattern to ensure uninterrupted access to report downloads.
- **If you use the dashboard UI**: No action is required if you download reports through the GitHub dashboard interface—the change is transparent to these users.

Refer to the [GitHub Copilot allowlist reference documentation](https://docs.github.com/copilot/reference/copilot-allowlist-reference) for a complete list of domains to allowlist. For questions or discussions, join the conversation in [GitHub Community Announcements](https://github.com/orgs/community/discussions/categories/announcements).

---

### Deeplink to AI coding tools
- Published: 2026-02-26T17:42:23+00:00
- Products: Linear
- Tags: feature, api, integration
- URL: https://standarddb.com/linear/linear-adds-deeplinks-to-ai-coding-tools-enables-prefilled-prompts-with-issue--vsfngdzg
- Original: https://linear.app/changelog/2026-02-26-deeplink-to-ai-coding-tools

## Deeplink to AI Coding Tools

Linear has streamlined the workflow for handing off work to coding agents. Previously, starting an issue meant manually creating a feature branch and copying relevant context. Now developers can launch their preferred coding tool directly from a Linear issue with a single keystroke.

**How it works:**
- Press `Ctrl/Cmd` + `Shift` + `;` to open your most recently used tool
- Press `W` then `O` to choose from enabled tools
- Click the button next to the issue identifier
- The coding tool receives a prefilled prompt including the issue ID, description, comments, updates, linked references, and images—no manual copying required

**Supported tools:**
Claude Code, Codex, Conductor, Cursor, GitHub Copilot, OpenCode, Replit, v0, and Zed.

**Customizable prompt templates:** Teams can define standing instructions for how agents should approach issues. For example, you might configure the template to require agents to provide a detailed plan before writing code, treat issue comments as part of the spec, or outline required tests before marking tasks complete. Configure personal preferences and prompt templates in the account settings.

## Additional Updates

Linear also shipped several improvements this release:

- **Notion integration:** Create and update Linear issues directly from Notion's new Custom Agents
- **API enhancements:** Markdown file links now render correctly in mutations; project webhooks trigger on milestone and relation changes; `diffMarkdown` field added to update payloads
- **Bug fixes:** Resolved issues with Mermaid diagram editing, Insights fullscreen navigation, SLA history rendering, project label management, and webhook duplicates
- **Performance improvements:** Enhanced emoji search for organizations with many custom emojis; agent session thoughts now render as full rich text
- **Keyboard shortcut fix:** The `/` key now correctly types as a character in action menus when text is present

---

### Nano Banana 2: Combining Pro capabilities with lightning-fast speed
- Published: 2026-02-26T16:01:50+00:00
- Products: Gemini, Google
- Tags: release, feature, model, api
- URL: https://standarddb.com/google/google-launches-nano-banana-2-image-model-combining-pro-capabilities-with-flash--znqxfhmp
- Original: https://deepmind.google/blog/nano-banana-2-combining-pro-capabilities-with-lightning-fast-speed

## Nano Banana 2 Now Available

Google DeepMind has launched **Nano Banana 2** (Gemini 3.1 Flash Image), a new state-of-the-art image generation model that merges the capabilities of Nano Banana Pro with the speed of Gemini Flash. This latest iteration makes advanced image generation features accessible to a broader audience by delivering production-ready quality at significantly faster speeds.

## Key Capabilities

Nano Banana 2 brings several powerful features to image generation:

- **Advanced world knowledge:** The model leverages Gemini's real-world knowledge base combined with real-time web search information to accurately render specific subjects and create detailed infographics, diagrams, and data visualizations.
- **Precision text rendering and translation:** Users can now generate accurate, legible text for marketing mockups and greeting cards, with built-in translation and localization capabilities for global sharing.
- **Subject consistency:** Enhanced control over visual elements and creative direction.
- **Lightning-fast inference:** Enables rapid edits and iterative workflows that were previously only available with slower models.

## Rollout and Availability

Nano Banana 2 is rolling out across Google's product ecosystem, including the Gemini app, Google Search, and Google Ads. This broader availability democratizes access to high-quality image generation previously limited to Pro-tier users.

## SynthID Content Credentials

Google is also advancing its AI-generated content identification capabilities by improving **SynthID** technology with **C2PA Content Credentials**, helping creators and platforms verify and track the origins of AI-generated images.

---

### Copilot Content Exclusion REST API in public preview
- Published: 2026-02-26T15:37:48+00:00
- Products: GitHub Copilot, GitHub Dependabot
- Tags: api, feature, release
- URL: https://standarddb.com/github/github-launches-copilot-content-exclusion-rest-api-in-public-preview--qjcvzkjw
- Original: https://github.blog/changelog/2026-02-26-copilot-content-exclusion-rest-api-in-public-preview

## New Content Exclusion REST API

GitHub has introduced a new REST API for managing Copilot content exclusion rules, now available in public preview. This JSON-based API enables organization and enterprise administrators to programmatically control which content is excluded from GitHub Copilot, automating what was previously a manual process.

## Key Capabilities

The API supports essential operations for content exclusion governance:

- **GET operations** to retrieve current exclusion rules at the organization and enterprise levels
- **SET operations** to define and update content exclusion policies
- Scope management at both organization and enterprise tiers
- RESTful endpoints aligned with GitHub's standard API design

## For Developers and Admins

Organization and enterprise administrators can now integrate content exclusion management into their existing automation workflows and governance tools. This API simplifies compliance and security practices by enabling programmatic policy enforcement at scale, rather than relying on manual configuration through the UI.

**Getting started:** Review the [API documentation](https://docs.github.com/enterprise-cloud@latest/rest/copilot/copilot-content-exclusion-management?apiVersion=2022-11-28) and [content exclusion concepts guide](https://docs.github.com/copilot/concepts/context/content-exclusion) to understand available endpoints and implementation patterns. Feedback is welcome in the [GitHub Community discussions](https://github.com/orgs/community/discussions/categories/announcements).

---

### New dashboard redesign is now the default
- Published: 2026-02-26T13:00:00+00:00
- Products: Vercel
- Tags: release, feature, platform
- URL: https://standarddb.com/vercel/vercel-rolls-out-redesigned-dashboard-with-sidebar-navigation-and-mobile--qhgkdmzc
- Original: https://vercel.com/changelog/dashboard-navigation-redesign-rollout

## Dashboard Redesign Goes Live

Vercel has completed its rollout of the new dashboard redesign as of February 26, 2026. After a successful opt-in beta period in January, the updated interface is now available to all users by default. The redesign was informed by user feedback collected during the beta phase.

## Key Features

The new dashboard navigation includes:

- **Resizable sidebar**: Horizontal tabs have been consolidated into a collapsible sidebar that users can hide when not needed
- **Unified navigation**: Consistent tab structure across both team and project levels for a cohesive experience
- **Prioritized workflows**: Navigation items are now ordered based on the most common developer tasks
- **Project-level filtering**: Users can switch between team and project views of the same page with a single click
- **Mobile-optimized interface**: A floating bottom bar design optimized for one-handed mobile navigation

## Getting Started

No action is required—the new navigation is automatically available to all Vercel users. Users can visit their [dashboard](https://vercel.com/dashboard) to start using the updated experience immediately.

---

### Nano Banana 2 is live on AI Gateway
- Published: 2026-02-26T13:00:00+00:00
- Products: Vercel AI SDK, Vercel
- Tags: release, feature, model, integration
- URL: https://standarddb.com/vercel/vercel-ai-gateway-adds-gemini-3-1-flash-image-preview-with-web-search-and--ytdgqnmz
- Original: https://vercel.com/changelog/nano-banana-2-is-live-on-ai-gateway

## Gemini 3.1 Flash Image Preview Now Available

Vercel has released Nano Banana 2 (Gemini 3.1 Flash Image Preview) on AI Gateway, a multimodal model that generates both images and text while maintaining the speed and cost efficiency of flash-tier models.

### Key Features

**Visual Grounding with Google Image Search**: The model can now retrieve live visual data from Google Image Search to ground outputs in real-world imagery. This is particularly useful for rendering lesser-known landmarks and objects with greater accuracy.

**Configurable Thinking Levels**: Developers can now set thinking levels to either "Minimal" or "High" to control how thoroughly the model reasons through complex prompts before generating responses.

**Expanded Format Support**: New resolutions and aspect ratios are available, including 512p images with 1:4 and 1:8 aspect ratios, enabling support for more types of creative assets beyond standard dimensions.

### How to Use

To integrate Nano Banana 2, set the model to `google/gemini-3.1-flash-image-preview` in the Vercel AI SDK and use `streamText` or `generateText` functions. The model supports multimodal responses by configuring `responseModalities` to include both TEXT and IMAGE. For advanced reasoning, enable `thinkingConfig` with `includeThoughts: true` and set `thinkingLevel` to 'high'.

### Additional Resources

AI Gateway provides a unified API for model access with built-in observability, Bring Your Own Key support, intelligent provider routing, automatic retries, and failover capabilities. Developers can explore available models on the [AI Gateway model leaderboard](https://vercel.com/ai-gateway/leaderboards) or test the model in Vercel's [model playground](https://vercel.com/ai-gateway/models/gemini-3.1-flash-image-preview).

---

### GitHub Mobile: Track coding agent progress in real time with Live Notifications
- Published: 2026-02-26T10:58:22+00:00
- Products: GitHub Copilot, GitHub
- Tags: feature, platform
- URL: https://standarddb.com/github/github-mobile-adds-live-notifications-for-copilot-coding-agent-sessions--gvfykjgk
- Original: https://github.blog/changelog/2026-02-26-github-mobile-track-coding-agent-progress-in-real-time-with-live-notifications

## Live Coding Agent Notifications Now Available

GitHub Mobile now includes **Live Activities** that surface real-time progress updates from coding agent sessions directly on your phone. Whether you initiated an agent session from GitHub.com, GitHub Mobile, or the CLI, you'll receive live notifications tracking the agent's work as it happens.

## What You Get

The Live Activities display key session information including:
- **Related pull request title** and current status
- **Session state**: In progress, Completed, Failed, or Cancelled
- **Quick access** to tap through to the pull request and review changes in-app

This eliminates the need to constantly check your desktop to stay informed about your coding agents' progress while on the go.

## Supported Platforms and Agents

**Compatible with:**
- **Copilot coding agent**
- Supported third-party agents (Codex, Claude)
- Cloud-run sessions only (on-device sessions not yet supported)

**Device requirements:**
- iOS 17.2 or later (for Live Activities)
- Android 16 or later (for Live Update Notifications)
- Note: Older Android versions will still receive standard progress notifications

## Managing Live Notifications

You have full control over the feature—disable Live Activities anytime in Settings, or tap the "Disable Live Updates" button on your first live notification for quick opt-out. This feature is available now in the latest production builds of [GitHub Mobile for iOS](https://apps.apple.com/us/app/github/id1477376905) and [Android](https://play.google.com/store/apps/details?id=com.github.android).

---

### Improved search on the issues dashboard
- Published: 2026-02-26T09:29:08+00:00
- Products: GitHub
- Tags: feature, api, platform
- URL: https://standarddb.com/github/github-brings-semantic-search-to-issues-dashboard-enabling-natural-language--fynykhcm
- Original: https://github.blog/changelog/2026-02-26-improved-search-on-the-issues-dashboard

## Semantic Search on Issues Dashboard

GitHub is expanding its semantic search capabilities to the Issues dashboard, bringing the same natural language search experience previously launched in January to a wider audience. Starting this week, the feature enters public preview, allowing developers to search for issues across multiple repositories using plain English queries rather than structured syntax.

### Key Capabilities and Limitations

The semantic search on the dashboard includes several important details for users:

- **Broad search scope**: Search across all your repositories, though cross-repository searches are limited to your top 100 repositories
- **Smart fallback**: Filter-only searches and exact-match queries (like those using quotation marks) automatically use lexical search
- **Easy switching**: A preview banner appears on the Issues dashboard with an option to revert to classic search at any time

### Additional GitHub Issues and Projects Improvements

Beyond semantic search, GitHub shipped several smaller improvements in the same release:

- **Permission-aware UI**: Label and milestone pickers no longer show "Create new" options to users without creation permissions
- **Better loading feedback**: Creating sub-issues now displays a loading state while the form loads
- **Agent improvements**: Fixed a bug where assigning third-party agents to issues didn't immediately trigger work—agents now start as expected
- **Performance boost**: GitHub Projects no longer shows item counts in the top navigation bar to improve page load times site-wide

### Getting Started

Visit the Issues dashboard to see the semantic search preview banner. You can provide feedback through the [GitHub Community discussions](https://github.com/orgs/community/discussions/188029).

---

### Bun v1.3.10
- Published: 2026-02-26T06:31:59+00:00
- Products: 
- Tags: release, feature, platform
- URL: https://standarddb.com/bun/bun-v1-3-10-launches-native-repl-adds-es-decorators-support-enables-self--tczmqnnn
- Original: https://bun.com/blog/bun-v1.3.10

## New Native REPL

Bun's REPL has been completely rewritten in Zig, replacing the previous third-party npm package implementation. The new REPL starts instantly without downloading any dependencies and includes a full-featured terminal UI with practical developer conveniences.

**Key features include:**
- **Copy to clipboard** — The `.copy` command copies expressions directly to your clipboard
- **Top-level await** — Use async/await natively in the REPL
- **Module loading** — Both ESM imports and CommonJS require work seamlessly
- **Syntax highlighting** — JavaScript code is colorized as you type
- **Line editing** — Emacs keybindings (`Ctrl+A/E`, `Ctrl+K/U`, `Ctrl+W`, `Ctrl+L`) for efficient navigation
- **Persistent history** — Command history saved to `~/.bun_repl_history` with Up/Down arrow navigation
- **Tab completion** — Auto-complete for object properties and REPL commands
- **Multi-line input** — Automatic detection of incomplete expressions with proper continuation

Special variables `_` (last result) and `_error` (last error) are available, and REPL semantics properly handle `const`/`let` hoisting, dynamic imports, and object literal detection.

## ES Decorators and Browser Compilation

Bun now fully supports **TC39 stage-3 standard ES decorators** — the modern decorator specification used when `experimentalDecorators` is disabled in `tsconfig.json`. This addresses a long-standing feature request dating to 2023. The transpiler now correctly handles decorator metadata via `Symbol.metadata`, the `accessor` keyword, and context APIs like `ClassMethodDecoratorContext` and `ClassFieldDecoratorContext`.

The `--compile --target=browser` flag produces self-contained HTML files with all JavaScript, CSS, and assets inlined directly into the output:
- Supports TypeScript, JSX, React, CSS, ESM, and CJS
- Useful for distributing `.html` files that work via `file://` URLs without web servers or CORS concerns
- Converts `<script src="...">` to inline `<script>`, `<link rel="stylesheet">` to `<style>`, and asset references to `data:` URIs

**Usage:**
```bash
bun build --compile --target=browser ./index.html
```

All entrypoints must be `.html` files and cannot use `--splitting`.

---

### OpenAI Codex and Figma launch seamless code-to-design experience
- Published: 2026-02-26T06:00:00+00:00
- Products: OpenAI API, OpenAI
- Tags: integration, feature, api
- URL: https://standarddb.com/openai/openai-codex-integrates-with-figma-enabling-seamless-code-to-design-workflows--sxlhfctm
- Original: https://openai.com/index/figma-partnership

## New Code-to-Design Integration

OpenAI and Figma have announced a new integration connecting Codex directly to Figma's design platform. This enables product builders to move seamlessly between code and design canvas, allowing teams to iterate faster and ship products with greater efficiency.

The integration leverages **Figma MCP Server**, an open-source standard that enables AI agents to interface with external applications and tools. Through this connection, users can:

- **Convert code to design**: Transform UI implementations from code into editable Figma designs for visual exploration and iteration
- **Convert design to code**: Bring Figma designs, Figma Make, and FigJam details directly into Codex for code implementation
- **Roundtrip workflows**: Start from a prompt, code, or design—and move between contexts without losing implementation details

## Expanding the Partnership

This release builds on the existing partnership between OpenAI and Figma. Figma already uses OpenAI models across its platform and was an early ChatGPT app partner in 2025. The code-to-design integration deepens this collaboration by bridging the gap between creative design and engineering workflows.

According to Figma's Chief Design Officer Loredana Crisan, this integration shifts focus from "whether you can build" to "what you build and how it stands out," enabling teams to iterate on their best ideas rather than their first ones.

## Getting Started

To use the integration, developers can install the Figma MCP Server directly within the Codex desktop application. The integration works with Codex across all platforms: the CLI, IDE extensions, web, and the Codex desktop app for macOS.

This comes as Codex adoption continues to accelerate, with over 1 million weekly users and 400% usage growth since the start of 2026. Enterprises including Cisco, NVIDIA, Ramp, and Datadog have already adopted Codex.

---

### v0.16.0
- Published: 2026-02-26T04:35:50+00:00
- Products: 
- Tags: release, feature, performance, api, sdk
- URL: https://standarddb.com/vllm/vllm-v0-16-0-brings-async-scheduling-with-pipeline-parallelism-30-8-throughput--gjbzpzbk
- Original: https://github.com/vllm-project/vllm/releases/tag/v0.16.0

## Core Performance Improvements

vLLM v0.16.0 marks a major release with 440 commits from 203 contributors, featuring **async scheduling + pipeline parallelism** now fully supported with documented **30.8% end-to-end throughput improvement** and **31.8% time-per-output-token improvement**. This represents a significant step forward for serving large language models at scale.

## New Capabilities

**Realtime API**: A new WebSocket-based Realtime API enables streaming audio interactions, building on the Voxtral realtime infrastructure for conversational AI applications.

**Speculative Decoding Enhancements**: 
- Unified Parallel Drafting framework simplifies deployment
- Speculative decoding now works with structured outputs
- Penalty application integrated into Model Runner V2
- Additional drafting model support including EAGLE3, AFMoE, and Mistral3

## Model Support Expansion

This release adds support for 12+ new model architectures including:
- Vision models: GLM-OCR with MTP, DeepSeek-OCR-2, Intern-S1-Pro, MiniCPM-o 4.5
- Speech/Audio: Qwen3-ASR, FunAudioChat
- Other specialized models: ColBERT, voyage-4-nano, GLM-5

Additionally, LoRA support expanded to Gemma3 vision components and Nemotron-H MTP models with optimized fused MoE-LoRA kernel indexing.

## Infrastructure & RLHF Improvements

- Native NCCL-based weight syncing API for distributed training
- Layerwise weight reloading for QeRL
- Engine pause/resume with request preservation for flexible training workflows
- **Major XPU platform overhaul**: Deprecated IPEX in favor of vllm-xpu-kernels with new support for MoE, MXFP4 MoE, WNA16, scaled_mm, and FP8 MoE

## Action Items

Users should update to access performance gains with async scheduling and pipeline parallelism. Teams using RLHF workflows or Intel XPU hardware should review updated APIs and deprecation notices.

---

### Org-level metrics API now includes pull request throughput metric parity
- Published: 2026-02-26T00:11:37+00:00
- Products: GitHub Copilot, GitHub Dependabot
- Tags: api, feature, platform
- URL: https://standarddb.com/github/github-expands-copilot-usage-metrics-api-with-org-level-pull-request-metrics--zvmycjtn
- Original: https://github.blog/changelog/2026-02-25-org-level-metrics-api-now-includes-pull-request-throughput-metric-parity

## Overview

GitHub is expanding the organization-level [Copilot Usage Metrics API](https://docs.github.com/enterprise-cloud@latest/rest/copilot/copilot-usage-metrics?apiVersion=2022-11-28) to include a broader set of pull request activity metrics. This update brings org-level endpoints to **feature parity** with enterprise-level capabilities that were previously exclusive to enterprise tier users.

## New Metrics Available

Organizations can now retrieve daily aggregated pull request activity through the org-level API:

**Baseline Pull Request Activity:**
- Total pull requests created, reviewed, and merged
- Median time to merge (in minutes)

**Copilot Participation Metrics:**
- Pull requests created by Copilot coding agent
- Pull requests reviewed by Copilot code review
- Review suggestions and applied suggestions from Copilot
- Copilot-created pull requests that were merged
- Median time to merge for Copilot-created pull requests

## Key Benefits

With these metrics now available at the org level, teams can:

- **Standardize reporting** across the organization without requiring enterprise-level access
- **Quantify Copilot's impact** on the entire pull request lifecycle—from creation through merge
- **Track actionable feedback** by measuring applied suggestions alongside total suggestions
- **Gain visibility** into non-IDE Copilot usage, including agent-created pull requests and Copilot-authored reviews

## Important Considerations

- **Deduplication differences**: If an organization belongs to an enterprise, org-level totals may not match enterprise totals due to users belonging to multiple organizations
- **Attribution edge cases**: Repository transfers can split attribution between organizations if transfers occur between pull request creation and merge
- **Data availability**: If the API reports contain only pull request data, it indicates the organization lacks IDE usage activity

For complete field definitions and implementation details, see the [Copilot usage metrics documentation](https://docs.github.com/copilot/reference/copilot-usage-metrics/copilot-usage-metrics).

---

### Dashboard enhancements and improvements
- Published: 2026-02-26T00:00:00+00:00
- Products: 
- Tags: release, feature, platform
- URL: https://standarddb.com/databricks/databricks-ai-bi-adds-agent-mode-pivot-table-hierarchies-and-chart-faceting-in--yvtjlzgc
- Original: https://docs.databricks.com/aws/en/ai-bi/release-notes/2026

## Dashboard Visualization and Interaction

Databricks enhanced dashboard capabilities with several visualization improvements. **Pivot tables** now support expanding and collapsing hierarchies, while **stacked bar charts** can display min/max label values and totals. **Area and line charts** gained the ability to sort axes by measures. Charts now support **faceting**, which renders one chart per dimension value for clearer data segmentation.

## Table and Layout Improvements

Table visualizations received significant upgrades including support for **aggregations and measures**, conditional formatting with color gradients, and background color styling. The dashboard canvas now supports **12-column layouts** instead of the previous 6-column limit, giving designers more flexibility for widget placement. Users can also resize table and pivot table headers and columns, with column widths set via the Style tab. **Pivot tables** now support cross-filtering and HTML, JSON, and image cell styles.

## Genie Agent Mode and AI Features

Databricks renamed "Research Agent" to **Agent mode (Beta)**, which now streams thinking traces inline and runs queries in parallel for faster results. **Inspect mode (Beta)** was introduced to automatically improve Genie accuracy by reviewing generated SQL and authoring verification queries. The table limit for Genie spaces increased from 25 to 30 tables. Genie now generates **heatmap and sankey visualizations**, supports metric view display names and formats, and automatically creates JOIN relationships based on primary and foreign keys.

## Filters, Parameters, and Admin Features

Date range parameters now support an **"All" selection**, returning `null` when selected to display all rows. Multi-select parameters gained the ability to select all values at once. A new **DD-MM-YYYY date format** is available for date fields. Dashboard refresh cancellation now displays the last available data instead of an error. The platform fixed issues where datetime drop-down menus couldn't be set in global filters and resolved a bug preventing workspace admins from editing workspace-level themes.

## Additional Capabilities

Unity Catalog metric view lineage now appears in dashboard info sections. Global search in Lakehouse and Databricks One now matches on dashboard content, not just titles. **Point maps** recenter upon relevant points when filters are applied. The Assistant for dashboard authoring expanded to support creating labels and setting color encoding values. **Level of detail calculations** now support `EXCLUDE` syntax on multiple fields, and color legends can be sorted by another field.


---

### Bugbot Autofix
- Published: 2026-02-26T00:00:00+00:00
- Products: 
- Tags: feature, api, integration
- URL: https://standarddb.com/cursor/cursor-s-bugbot-autofix-now-proposes-pull-request-fixes-automatically-35-merge--vjhgyzsy
- Original: https://cursor.com/changelog/02-26-26

## Bugbot Autofix Now Available

Cursor has introduced **Bugbot Autofix**, a new capability that automatically detects and fixes issues found in pull requests. The feature leverages cloud-based agents that test changes on their own machines and propose fixes directly within your PR workflow.

### How It Works

Bugbot Autofix operates in two modes:

- **Review Mode**: Posts a comment on the original PR with a preview of proposed fixes. Developers can merge changes using a provided `@cursor` command for manual control.
- **Auto-Push Mode**: Configured to directly push changes to your branch with no interaction required, streamlining the workflow for teams that prefer hands-off automation.

### Early Adoption Success

Early results are encouraging—over 35% of proposed Bugbot Autofix changes are already being merged into base PRs, indicating strong adoption and confidence in the tool's accuracy.

### Getting Started

To enable this feature, navigate to your [Bugbot dashboard](https://cursor.com/dashboard?tab=bugbot) and configure your preferred autofix settings. For more details, see the [Bugbot Autofix documentation](https://cursor.com/docs/bugbot#autofix).

---

### Enterprise AI Controls & agent control plane now generally available
- Published: 2026-02-26T00:00:00+00:00
- Products: GitHub Copilot, GitHub Dependabot
- Tags: release, feature, api, platform
- URL: https://standarddb.com/github/github-s-enterprise-ai-controls-reaches-general-availability-with-agent--phcllksn
- Original: https://github.blog/changelog/2026-02-26-enterprise-ai-controls-agent-control-plane-now-generally-available

## Overview

GitHub's **Enterprise AI Controls and agent control plane** are now generally available, providing enterprise administrators with comprehensive governance tools for managing AI agents and Copilot usage at scale. These features address the core challenge of enabling AI capabilities while maintaining security, compliance, and visibility across the organization.

## Key Features

**AI Administrator Role & Workspace**: A new enterprise custom role with fine-grained permissions allows designated AI administrators to oversee all AI-related settings from a centralized console. This role grants access to audit logs, agent session activity, and AI Controls management without requiring full enterprise admin access.

**Complete Agent Activity Visibility**: The audit logging system now provides full visibility into agent operations, including:
- `actor_is_agent` identifiers that distinguish agent-driven actions
- User attribution showing who agents act on behalf of
- `agent_session.task` events tracking session lifecycle (started, finished, failed)
- Cloud agent session activity from the past 24 hours
- Centralized MCP (Model Context Protocol) registry and allowlist management

**Custom Agent Standards & Protection**: Organizations can now define, version control, and enforce enterprise standards for custom agents. A new 1-click push rule protects the `.github/agents/*.md` file path across the enterprise, preventing unauthorized modifications while allowing controlled updates.

## New in General Availability

**Enhanced Agent Discovery**: Administrators can now search and filter agentic session activity by specific agents (including third-party agents) and track usage by organization. The agents page audit log is prefiltered for Copilot and third-party agents, and the 24-hour session history no longer caps at 1,000 records, enabling complete session traceability.

**Enterprise Agent Policies via API**: New REST API endpoints allow programmatic application of enterprise-wide custom agent definitions, enabling Infrastructure-as-Code approaches to agent governance. The **AI Controls** tab in enterprise settings is now the permanent home for all AI-related policies and settings.

**MCP Allow Lists in Preview**: Enterprise-wide MCP allow list management remains in public preview while GitHub refines the solution to scale better across organizations without requiring multiple registries.

## What Developers & Admins Should Know

- Enterprise administrators should navigate to the new **AI Controls** tab in enterprise settings to configure agent policies and review activity
- Custom agent definitions are now stored in `.github-private/agents/*.md` for canonical control
- API support enables programmatic policy management for enterprise-scale deployments
- MCP governance features are still evolving; additional session activity coverage and granular policy controls are planned

## Roadmap

Future enhancements include comprehensive session activity coverage across VS Code and Copilot CLI clients, programmatic access to agent activity data, and expanded MCP governance options.

---

### Developer role now available for Pro teams
- Published: 2026-02-26T00:00:00+00:00
- Products: Vercel
- Tags: feature, platform
- URL: https://standarddb.com/vercel/vercel-expands-developer-role-to-pro-teams-enabling-granular-access-control--bxxmmqdl
- Original: https://vercel.com/changelog/developer-role-now-available-for-pro-teams

## Developer Role Now Available for Pro Teams

Vercel has extended the Developer role to Pro tier teams, democratizing role-based access control (RBAC) that was previously limited to Enterprise customers. This change enables Pro team owners to assign more fine-grained permissions to team members.

### What's New

- **Developer role for Pro teams**: Owners can now assign the Developer role to existing team members or new invitees
- **Limited permissions**: Developers can deploy to projects but have restricted access to team-wide configuration and environment variables
- **Safer deployments**: Reduces the risk of accidental infrastructure changes by limiting developer access to deployment-specific tasks

### For Team Owners

Assign the Developer role through your team members settings in Vercel. This allows you to give developers production deployment capabilities while keeping sensitive configuration and organization settings protected.

### Learn More

See the [team level roles documentation](https://vercel.com/docs/rbac/access-roles/team-level-roles) for detailed permission matrices and role definitions.

---

### Cloudflare Fundamentals - Markdown responses for Cloudflare 1xxx errors
- Published: 2026-02-26T00:00:00+00:00
- Products: Cloudflare
- Tags: feature, api, performance
- URL: https://standarddb.com/cloudflare/cloudflare-adds-markdown-error-responses-for-1xxx-errors-reduces-payload-size--ptykxnzq
- Original: https://developers.cloudflare.com/changelog/post/2026-02-26-markdown-responses-for-1xxx-errors

## Markdown Error Responses Now Available

Cloudflare has introduced support for structured Markdown responses for Cloudflare-generated 1xxx errors. Clients can opt-in by sending `Accept: text/markdown` in their request headers, enabling more efficient error handling and parsing for automated systems.

## Response Format and Metadata

Each Markdown response includes comprehensive structured data via YAML frontmatter and two guidance sections:

- **Frontmatter fields**: `error_code`, `error_name`, `error_category`, `http_status`, `ray_id`, `timestamp`, `zone`, `cloudflare_error`, `retryable`, `retry_after`, and `owner_action_required`
- **Guidance sections**: "What happened" and "What you should do" to enable deterministic retry and escalation decisions without HTML parsing

## Performance Impact

In measured tests across 1,015 comparisons, Markdown responses reduced payload size and token footprint by **over 98%** compared to HTML error pages. This significantly improves efficiency for agents and automated systems processing errors.

## Content Negotiation

Cloudflare uses standard HTTP content negotiation on the `Accept` header:

- `Accept: text/markdown` → Markdown response
- `Accept: text/markdown, text/html;q=0.9` → Markdown response
- `Accept: text/*` → Markdown response
- `Accept: */*` → HTML response (default)

When multiple content types are specified, Cloudflare selects the highest-priority supported media type using quality (`q`) values. If Markdown is not explicitly preferred, HTML is returned.

## Getting Started

To receive Markdown error responses, include the `Accept: text/markdown` header in your requests:

```bash
curl -H "Accept: text/markdown" https://<your-domain>/cdn-cgi/error/1015
```

Default behavior remains unchanged—clients that do not explicitly request Markdown continue to receive HTML error pages. Full documentation on Cloudflare 1xxx errors is available in the [error reference guide](https://developers.cloudflare.com/support/troubleshooting/http-status-codes/cloudflare-1xxx-errors/).

---

### Databricks Runtime maintenance updates (02/26)
- Published: 2026-02-26T00:00:00+00:00
- Products: 
- Tags: release, feature, platform, security, bugfix, integration, api
- URL: https://standarddb.com/databricks/databricks-adds-python-unit-testing-ismap-compliance-and-maintenance-updates--mzzmrjrz
- Original: https://docs.databricks.com/aws/en/release-notes/product/2026/february

## Key Features and Updates

Databricks shipped several significant platform improvements in February 2026:

### Testing and Development
The workspace now includes integrated Python unit testing capabilities, allowing developers to discover, run, and debug pytest-based tests directly within the IDE. The feature provides a dedicated tests sidebar, inline execution glyphs, and a testing results panel for streamlined test workflow.

### Compliance and Security
Two new compliance features are now available: **ISMAP compliance support** (generally available) enables workspaces to meet Japanese government cloud certification requirements. Additionally, **HITRUST compliance controls** entered public preview, helping organizations manage risk and demonstrate security and privacy compliance. These updates reflect Databricks' focus on regulated industry adoption.

### Runtime and Infrastructure
Maintenance updates are available across eight supported Databricks Runtime versions (12.2 through 18.0), delivering bug fixes, security patches, and performance improvements. **JAR tasks on serverless compute** moved to public preview, expanding serverless capabilities. The **Databricks ODBC Driver** was renamed from Simba Spark ODBC Driver, with existing versions supported for two years. **Serverless outbound IPs** are now retrievable via public JSON endpoint in preview.

### Data Integration and Models
**ADBC (Arrow Database Connectivity)** is now the default driver for new Power BI connections, improving performance and modern data access patterns. New foundation models are available through Mosaic AI Model Serving: **OpenAI GPT-5.3 Codex** for advanced code generation, and **Google Gemini 3.1 Pro Preview**. A new SQL feature allows adding and updating comments on streaming tables, now generally available.

---

### Claude and Codex now available for Copilot Business & Pro users
- Published: 2026-02-26T00:00:00+00:00
- Products: GitHub Copilot, GitHub
- Tags: feature, integration, platform, api
- URL: https://standarddb.com/github/github-expands-claude-and-codex-availability-to-copilot-business-and-pro-users--pgwbpgjx
- Original: https://github.blog/changelog/2026-02-26-claude-and-codex-now-available-for-copilot-business-pro-users

## Expanded Agent Availability

GitHub has expanded access to third-party AI coding agents, making Claude and Codex available to Copilot Business and Pro customers. This follows earlier availability for Enterprise and Pro+ tiers, bringing advanced agent-based code assistance to a broader developer audience at no additional cost.

## Unified Platform with Shared Context

All three agents—Claude, Codex, and GitHub's native Copilot—run on a single shared platform with **unified governance, shared memory, and shared context**. Agents have access to:

- Repository code and complete git history
- Issues and pull requests
- Copilot Memory for personalized context
- Repository-specific instructions and policies

This architecture ensures developers can seamlessly switch between agents while maintaining consistent context across web, mobile, and desktop environments.

## Expanded Capabilities Across Surfaces

Developers can now leverage these agents across multiple interfaces:

- **GitHub web**: Open the Agents tab in any enabled repository and choose an agent from the dropdown
- **GitHub Mobile**: Access agents from the dedicated "Agents" view or quick action menu
- **VS Code**: Launch agent sessions from the title bar (requires v1.109+) with options for local, cloud, or background execution
- **Issue/PR workflows**: Assign agents to issues, mention `@claude` or `@codex` in PR comments, or assign multiple agents to compare approaches

## Enterprise Controls and Administration

For Copilot Business customers, admins must enable agents at both enterprise and organization levels through the new **Enterprise AI Controls** interface. Copilot Pro users can self-serve by selecting repositories and toggling agent access. All activity runs within GitHub's audit logging and policy management systems via the Agent Control Plane, now generally available.

## Pricing and Request Consumption

Access to Claude and Codex is **fully included with existing Copilot subscriptions**—no additional fees required. Each coding agent session consumes one premium request during the public preview period.

---

### macos-26 is now generally available for GitHub-hosted runners
- Published: 2026-02-26T00:00:00+00:00
- Products: GitHub Actions
- Tags: release, feature, platform
- URL: https://standarddb.com/github/github-actions-macos-26-runners-now-generally-available-with-apple-silicon-and--nlbzsjsg
- Original: https://github.blog/changelog/2026-02-26-macos-26-is-now-generally-available-for-github-hosted-runners

## macOS 26 Runners Now Generally Available

GitHub Actions users can now access the macOS 26 runner image in production environments. Previously available as a public preview, macOS 26 runners are now fully supported and recommended for CI/CD workflows that require the latest macOS development tools.

## Runner Options and Labels

The macOS 26 image is available across four distinct runner configurations to suit different workload requirements:

- **`macos-26`** – Standard macOS runner optimized for arm64 (Apple Silicon)
- **`macos-26-intel`** – Standard macOS runner for x64 (Intel) architecture
- **`macos-26-large`** – Large runner designed for resource-intensive x64 builds
- **`macos-26-xlarge`** – Extra-large runner for demanding arm64 workloads

## What Developers Need to Know

Users can immediately adopt macOS 26 by updating the `runs-on` labels in their GitHub Actions workflow files. The image includes the latest macOS and Xcode tooling, making it suitable for modern app development and testing. For detailed information about included software and capabilities, the GitHub-hosted runners documentation provides comprehensive configuration guidance.

This release reinforces GitHub's commitment to supporting both Apple Silicon and Intel-based development workflows, allowing teams to choose the architecture that best fits their build requirements.

---

### Cache - Asynchronous stale-while-revalidate
- Published: 2026-02-26T00:00:00+00:00
- Products: Cloudflare
- Tags: feature, performance, api
- URL: https://standarddb.com/cloudflare/cloudflare-cache-now-revalidates-asynchronously-reducing-latency-and-error--xcspxxbz
- Original: https://developers.cloudflare.com/changelog/post/2026-02-26-async-stale-while-revalidate

## Asynchronous Stale-While-Revalidate Now Live

Cloudflare's `stale-while-revalidate` cache feature now operates fully asynchronously, eliminating the need for any visitor to wait for origin responses. This represents a significant improvement to cache behavior for sites using the `Cache-Control` directive.

### What Changed

Previously, the first request for an expired cached asset would block until the origin server responded. Now:
- **First request after expiration**: Triggers background revalidation and immediately returns stale content with `UPDATING` status
- **Subsequent requests during revalidation**: Also receive stale content with `UPDATING` status from cache
- **After origin responds**: All following requests receive fresh content with `HIT` status

### Key Benefits

- **Lower latency**: Every visitor gets served from cache immediately, with no origin wait times
- **Consistent experience**: All users receive identical cached responses during revalidation
- **Reduced error exposure**: The first request is no longer vulnerable to origin timeouts or errors—cached content is always returned while revalidation happens silently

### Deployment Status

The feature is live for all Free, Pro, and Business plan zones. Approximately 75% of Enterprise zones have been migrated, with remaining Enterprise deployments rolling out throughout Q1 2026.

### How to Use

Ensure your origin includes the `stale-while-revalidate` directive in the `Cache-Control` header. For example: `Cache-Control: max-age=3600, stale-while-revalidate=86400`. Refer to [Cloudflare's Cache-Control documentation](https://developers.cloudflare.com/cache/concepts/cache-control/#revalidation) for full details.

---

### Generate pull request titles with Copilot on the web
- Published: 2026-02-25T23:51:39+00:00
- Products: GitHub Copilot, GitHub
- Tags: feature, release
- URL: https://standarddb.com/github/github-copilot-now-auto-generates-pull-request-titles-from-commit-messages-on--hjfxzsxb
- Original: https://github.blog/changelog/2026-02-25-generate-pull-request-titles-with-copilot-on-the-web

## New Feature: AI-Powered PR Title Generation

GitHub has rolled out a new capability for GitHub Copilot that generates pull request titles directly on github.com. When creating or editing a PR, users will see a **GitHub Copilot** button in the title field that suggests descriptive titles based on the commit messages included in the pull request.

## How It Works

The feature integrates seamlessly into the PR creation workflow:
- Open a pull request on github.com
- Click the GitHub Copilot button in the title field
- Review the AI-generated suggestion and apply or edit it as needed

## Getting the Best Results

The quality of generated titles depends directly on commit message quality:
- **Use clear, descriptive commit messages** that explain what changed
- **Avoid placeholder messages** like `WIP` or generic `fix` without context
- **Group related changes** into a single PR so the title reflects a cohesive theme

If your commit messages are vague or repetitive, Copilot's suggestions may be less useful. This design incentivizes developers to maintain better commit hygiene.

## Availability

This feature is now **generally available** on github.com for any pull request containing one or more commits. Users can provide feedback in the [GitHub Community discussions](https://github.com/orgs/community/discussions/187249).

---

### GPT-5.3-Codex is now available in github.com, GitHub Mobile, and Visual Studio
- Published: 2026-02-25T23:25:30+00:00
- Products: GitHub Copilot, GitHub
- Tags: model, feature, release
- URL: https://standarddb.com/github/github-launches-gpt-5-3-codex-in-copilot-across-web-mobile-and-vs--fcsdybgy
- Original: https://github.blog/changelog/2026-02-25-gpt-5-3-codex-is-now-available-in-github-com-github-mobile-and-visual-studio

## Availability

GitHub Copilot now supports the **GPT-5.3-Codex** model, available to all subscription tiers: Copilot Enterprise, Copilot Business, Copilot Pro, and Copilot Pro+. Users can access this new model through the chat model picker across multiple platforms.

## Where to Access

The GPT-5.3-Codex model is available in:
- **GitHub.com** – Copilot Chat on the web
- **GitHub Mobile** – Native mobile app
- **Visual Studio Code** – Editor integration
- **Visual Studio** – Full IDE support

## How to Enable

For Copilot Business and Enterprise organizations:
- Administrators must opt in by enabling the **GPT-5.3-Codex policy** in Copilot settings
- Once enabled, the model will appear automatically in the model picker for all team members
- Individual Pro and Pro+ users get access immediately

## Next Steps

Users can review the [official documentation](https://docs.github.com/copilot/reference/ai-models/supported-models) to learn about all available Copilot models and start experimenting with GPT-5.3-Codex. Organizations can share feedback in the [GitHub Community forums](https://github.com/orgs/community/discussions/categories/copilot-conversations).

---

### Improved web search in Copilot on github.com
- Published: 2026-02-25T22:52:05+00:00
- Products: GitHub Copilot, GitHub Dependabot
- Tags: feature, release, platform
- URL: https://standarddb.com/github/github-copilot-gains-native-web-search-capabilities-for-real-time-answers--xrdjmhbl
- Original: https://github.blog/changelog/2026-02-25-improved-web-search-in-copilot-on-github-com

## What's New

GitHub Copilot Chat on github.com now supports **model native web search** for answering current events and time-sensitive questions. This enhancement provides faster and more accurate responses compared to previous search implementations.

## Supported Models

The feature is currently available on the following models:
- GPT-5.1
- GPT-5.1-Codex
- GPT-5.1-Codex-Mini (public preview)
- GPT-5.1-Codex-Max
- GPT-5.2-Codex

Other models will continue using Bing search as before.

## How to Access

- **Available to:** All paid Copilot subscribers with public preview features enabled
- **Organizations & Enterprises:** Enable via **Opt in to preview features** in Copilot settings on github.com
- **Opt-out:** Users can disable model native search by toggling off **"Copilot can search the web using model native search"** in their Copilot settings

## What Developers Should Know

This capability improves Copilot's ability to provide current information without relying on static training data. The gradual rollout across models allows GitHub to validate performance and gather feedback before broader deployment.

## What's Next

GitHub plans to expand this functionality across additional models and contexts throughout the Copilot platform. The development team is actively soliciting feedback through the GitHub community discussions.

---

### Announcing major updates to Procedures and Simulations: Enabling Fin to handle complex queries
- Published: 2026-02-25T20:00:13+00:00
- Products: 
- Tags: feature, release, api, platform
- URL: https://standarddb.com/intercom/intercom-launches-12-major-procedures-and-simulations-updates-for-complex-ai--pnqkzsrp
- Original: https://www.intercom.com/blog/procedures-simulations-updates

## Procedures: Natural Language Meets Deterministic Control

Intercom's Procedures enable teams to combine three powerful elements—natural language instructions, deterministic controls, and agentic behavior—to handle complex customer queries while maintaining predictability and control. The updates make building and maintaining Procedures significantly easier:

- **AI-Assisted Drafting**: Teams can outline a process and let AI draft complete Procedures based on conversation history, knowledge hub content, and relevant data, with the system prompting for clarifications as needed.
- **Sub-procedures**: Complex workflows can now be broken into reusable, modular steps, making workflows easier to read, faster to build, and simpler to maintain.
- **Improved Deterministic Controls**: New features include explicit Procedure switching under defined conditions, the ability to instruct Fin to reference specific knowledge hub content, and internal notes for seamless human handoffs.

## Enhanced Agentic Capabilities and Testing

The updates strengthen Fin's ability to handle real-world conversations where customers change topics or requirements mid-stream:

- **Automatic Procedure Switching**: Fin now intelligently detects when customer intent shifts and switches to the appropriate Procedure automatically without human intervention.
- **Data Extraction**: Fin can extract structured data directly from uploaded PDFs and images, reducing the need for customers to repeat information.

## Simulations: Confident Deployment at Scale

Simulations allow teams to test complex Procedures before reaching production:

- **AI-Suggested Test Cases**: Generate Simulations directly from Procedures to identify edge cases and potential issues.
- **Rich Testing**: Upload images for more realistic test scenarios and gain clearer visibility into Fin's reasoning.
- **The Fin Flywheel**: These updates support Intercom's train-test-deploy-analyze cycle, helping teams mature their AI workflows over time through continuous validation and refinement.

These updates address a key customer pain point: teams often restrict AI to simple tasks because they lack confidence in setting up and managing complex agentic behavior. With improved tools for defining logic without engineering involvement, Intercom aims to help teams unlock AI's full potential for high-value, complex work.

---

### GitHub Copilot CLI is now generally available
- Published: 2026-02-25T17:01:29+00:00
- Products: GitHub Copilot
- Tags: release, feature, api
- URL: https://standarddb.com/github/github-copilot-cli-reaches-general-availability-with-autonomous-agentic--mzthbykj
- Original: https://github.blog/changelog/2026-02-25-github-copilot-cli-is-now-generally-available

## GitHub Copilot CLI is Now Generally Available

GitHub Copilot CLI—a terminal-native AI coding agent—is now **generally available** for all paid GitHub Copilot subscribers. Since launching in public preview in September 2025, the tool has evolved from a simple terminal assistant into a full agentic development environment capable of planning complex tasks, executing workflows, editing files, running tests, and iterating autonomously.

### Agentic Development Capabilities

The core feature set distinguishes Copilot CLI from standard chat interfaces:

- **Plan Mode** (`Shift` + `Tab`): Copilot analyzes requests, asks clarifying questions, and builds a structured implementation plan before executing any code. Users can review and approve plans before execution begins.
- **Autopilot Mode**: For trusted tasks, Copilot operates fully autonomously—executing tools, running commands, and iterating without approval gates.
- **Specialized Agents**: The tool automatically delegates to domain-specific agents including Explore (codebase analysis), Task (builds and tests), Code Review (change analysis), and Plan (implementation planning). Multiple agents can run in parallel.
- **Background Delegation**: Use the `&` prefix to offload work to GitHub's cloud-based coding agent while keeping your local terminal free. Switch between local and remote sessions with `/resume`.

### Model Selection and Extensibility

Users can choose from the latest models from Anthropic (Claude Opus 4.6, Sonnet 4.6, Haiku 4.5), OpenAI (GPT-5.3-Codex, GPT-5 mini, GPT-4.1), and Google (Gemini 3 Pro). Switch models mid-session with `/model` and configure reasoning effort for extended-thinking models. GPT-5 mini and GPT-4.1 are included with Copilot subscriptions at no additional cost.

Extensibility is built-in through Model Context Protocol (MCP) servers, plugins, markdown-based skills, custom agents, and lifecycle hooks (`preToolUse` and `postToolUse`). Plugins can be installed directly from GitHub repositories using `/plugin install owner/repo`.

### Session Management and Code Review

New features include `/diff` for syntax-highlighted inline diffs with line-specific comments, `/review` for quick sanity checks before committing, and `Esc`–`Esc` to rewind file changes. Sessions support auto-compaction when approaching context limits and cross-session repository memory—Copilot remembers codebase conventions and patterns across sessions.

### Installation and Platform Support

Copilot CLI runs on macOS, Linux, and Windows via npm, Homebrew, WinGet, shell scripts, and standalone executables. Homebrew, WinGet, and script-based installations auto-update. The tool is included by default in GitHub Codespaces and available as a Dev Container Feature.

The terminal experience includes an experimental full-screen alt-screen mode with mouse support, customizable themes (GitHub Dark, Light, colorblind variants), UNIX keybindings (`Ctrl`+`A/E/W/U/K`), editor integration (`Ctrl`+`X`/`E`), and accessibility features including screen reader mode.

---

### v0.17.1-rc2
- Published: 2026-02-25T09:09:58+00:00
- Products: 
- Tags: 
- URL: https://standarddb.com/ollama/v0-17-1-rc2--spjlmsxc
- Original: https://github.com/ollama/ollama/releases/tag/v0.17.1-rc2



---

### v0.5.100
- Published: 2026-02-25T08:38:04+00:00
- Products: Sim Studio
- Tags: release, feature, integration, api, security, performance, bugfix
- URL: https://standarddb.com/sim/sim-v0-5-100-adds-attio-and-gong-integrations-credential-management-overhaul--ncqtqmtz
- Original: https://github.com/simstudioai/sim/releases/tag/v0.5.100

## New Integrations and Features

Sim v0.5.100 introduces two major integrations expanding enterprise platform support:

- **Attio CRM integration**: Adds 40 new tools and 18 webhook triggers, with automatic webhook lifecycle management
- **Gong integration**: Brings 18 API tools for revenue intelligence workflows
- **Confluence webhooks**: New webhook triggers enable event-driven automation from Confluence

## Credentials and Security

The release significantly enhances credential management with several security-focused improvements:

- **Multiple credentials per provider** now supported, enabling multi-tenant or multi-account scenarios
- **Public API access controls**: New environment variables and permission group settings allow fine-grained control over public API access
- **Improved credentials UI**: Enhanced user interface with bulk paste functionality and save notifications
- Credential management bug fixes for dependent endpoints

## API and Performance Enhancements

- **Retry configuration for API blocks**: Developers can now configure retry logic for API calls
- **Reduced database queries**: Execution preprocessing optimized to eliminate redundant DB queries
- **LLM SDK improvements**: Abort signals now properly propagate to all LLM provider calls
- **Redis connection handling**: Improved stale TCP connection detection with faster lease deadline handling

## Additional Improvements

- Audit logging enriched across 23 call sites for better observability
- Authentication fixes including DISABLE_AUTH support in web app
- Block naming preservation when pasting workflows without conflicts
- Improved tag dropdown to exclude invalid downstream blocks in loops and parallel siblings

---

### v0.16.1rc0: [ROCm][CI] Pin TorchCodec to v0.10.0 for ROCm compatibility (#34447)
- Published: 2026-02-25T04:30:22+00:00
- Products: 
- Tags: release, bugfix
- URL: https://standarddb.com/vllm/vllm-v0-16-1rc0-pins-torchcodec-to-v0-10-0-for-rocm-compatibility--fhxzqjwc
- Original: https://github.com/vllm-project/vllm/releases/tag/v0.16.1rc0

## vLLM v0.16.1rc0 Release

This is a release candidate for vLLM v0.16.1, focusing on improving compatibility with AMD's ROCm platform.

### What Changed

- **TorchCodec Pinned to v0.10.0**: The release pins TorchCodec dependency to version 0.10.0 to ensure proper ROCm compatibility. This addresses compatibility issues that could arise from using newer versions of TorchCodec with ROCm environments.

### Why This Matters

The fix was cherry-picked from previous commits to stabilize the ROCm continuous integration pipeline and ensure that users running vLLM on AMD GPUs get a consistent experience without dependency conflicts.

### For ROCm Users

If you're using vLLM with AMD GPUs and ROCm, upgrading to v0.16.1 when released will ensure you have the correct TorchCodec version pinned, avoiding potential compatibility issues.

### Release Status

This is a release candidate (rc0), so it's available for testing before the final v0.16.1 release.

---

### v0.17.1-rc1
- Published: 2026-02-25T04:08:05+00:00
- Products: 
- Tags: release, feature, model
- URL: https://standarddb.com/ollama/ollama-v0-17-1-rc1-adds-support-for-qwen3-5-architecture--sdvfmwxl
- Original: https://github.com/ollama/ollama/releases/tag/v0.17.1-rc1

## Ollama v0.17.1-rc1 Released

Ollama has released v0.17.1-rc1, a release candidate version that adds support for the **Qwen3.5 architecture**.

### What's New

- **Qwen3.5 Architecture Support**: The release introduces compatibility with Qwen3.5 models, expanding the range of models users can run locally through Ollama.

### Getting Started

Users can now pull and run Qwen3.5-based models using Ollama's standard commands once the release is finalized. As a release candidate, this version is available for testing before the stable release.

### Next Steps

Test the release candidate and report any issues to help stabilize v0.17.1 for general availability.

---

### February 24, 2026
- Published: 2026-02-25T01:01:55+00:00
- Products: 
- Tags: release, feature, api, performance
- URL: https://standarddb.com/devin/devin-2-2-launch-3x-faster-startup-desktop-testing-v3-api-exit-beta--gfqlhwqr
- Original: https://docs.devin.ai/release-notes/overview

## Devin 2.2 Major Release

Devin has released version 2.2, marking a significant milestone with hundreds of incremental improvements accumulated over recent weeks. The update focuses on improving developer workflows and performance across the platform.

### Key Performance Improvements

- **3x faster startup time**: Sessions now display results immediately, helping developers gain confidence that the agent is on the right track
- **Redesigned UI**: The new interface connects every step of the development lifecycle, allowing developers to start sessions from anywhere, review agent output directly in Devin, and jump back into sessions from code review
- **Improved integrations**: Slack and Linear integrations are now smoother and faster, enabling developers to start sessions without context switching

### Desktop Testing and Computer Use

Devin now supports full end-to-end testing using computer use capabilities for any desktop application that runs on Linux. The workflow is straightforward: when Devin creates a PR, it can request QA approval, run the application, use the desktop to interact with it, and provide an edited recording of the testing session for your review. Existing users can enable Desktop mode in Settings > Customization.

### v3 API Now Generally Available

The v3 API has officially exited beta and is now the primary API for all Devin functionality. It provides all legacy API capabilities plus role-based access control, session attribution, and new features. **Important**: Legacy APIs (v1 and v2) will be deprecated in the future with at least 30 days notice. All new features will be exclusive to the v3 API.

### Additional Improvements

- **Sessions List Redesign**: Updated layout with inline PR previews, message snippets, and status indicators; sessions can be sorted by creation date
- **Merge Conflict Detection**: Automatic notifications when PRs have merge conflicts (GitHub.com)
- **Session Messages API**: New `GET /messages` endpoint for programmatic access to session history
- **Devin Review for GitHub Enterprise Server**: Support for GHES repositories with diff viewing, analysis, and chat-based code changes
- **Scheduling Enhancements**: One-time scheduled events and "Run now" button for on-demand triggers

### Additional Features (Recent Updates)

- **Devin Fast Mode**: ~2x faster responses at 4x ACU per session
- **Code Changes from Chat**: Review chat agent can propose code edits directly in conversation
- **Batch Comments**: Start a review to batch multiple PR review comments
- **Skills Support**: Devin now recognizes and uses team-defined skills in your codebase
- **Secure Mode for All Organizations**: Now available to non-enterprise organizations

Developers should review the v3 API documentation and plan migration from v1/v2 APIs during the deprecation period.

---

### Workers - Better Windows support for Python Workers
- Published: 2026-02-25T00:00:00+00:00
- Products: Cloudflare Workers
- Tags: feature, sdk, platform
- URL: https://standarddb.com/cloudflare/cloudflare-extends-python-workers-to-windows-with-pywrangler-support--pwclgzpc
- Original: https://developers.cloudflare.com/changelog/post/2026-02-13-pywrangler-windows-support

## Windows Support for Python Workers

Cloudflare has extended Pywrangler, the CLI tool for Python Workers, to support Windows development environments. Previously limited to macOS and Linux, Python Workers can now be developed and deployed directly from Windows.

### Getting Started

To develop and deploy Python Workers on Windows, use Pywrangler via `uvx`:

```bash
uvx --from workers-py pywrangler dev
uvx --from workers-py pywrangler deploy
```

Specify your Worker's Python dependencies in `pyproject.toml` as usual—no additional Windows-specific configuration is needed.

### Requirements

To use this feature, ensure you have the minimum required versions:

- **wrangler** >= 4.64.0
- **workers-py** >= 1.72.0
- **uv** >= 0.29.8

Upgrade your tools with:
- `uv tool upgrade workers-py` (includes Pywrangler)
- `npm install -g wrangler@latest`
- `uv self update` (for uv)

### What's Included

All Pywrangler functionality works on Windows, including package management, local development, and deployment. This brings feature parity across all supported operating systems.

---

### Adds support for tax collection in Sri Lanka for digital services
- Published: 2026-02-25T00:00:00+00:00
- Products: Stripe API
- Tags: api, feature, release
- URL: https://standarddb.com/stripe/stripe-adds-tax-collection-support-for-sri-lanka-digital-services--xpdsmnmk
- Original: https://docs.stripe.com/changelog.md

Stripe has announced expanded tax support for merchants operating in Sri Lanka. The platform now provides tax collection capabilities specifically designed for digital services in this region, allowing businesses to comply with local tax regulations.

## Key Changes

- **Sri Lanka Digital Services Tax**: Adds support for tax collection specifically for digital services provided in Sri Lanka
- **VAT ID Support**: Introduces support for Sri Lanka VAT tax identification numbers, enabling proper tax registration and compliance tracking

## What Developers Need to Know

This update is part of Stripe's ongoing effort to expand tax compliance coverage globally. Merchants and platforms using Stripe's Tax API can now configure tax settings for Sri Lankan digital services, ensuring they collect and remit taxes in compliance with local regulations.

The changes are non-breaking and available immediately through the Stripe API. Developers should consult the updated Tax API documentation to implement the new Sri Lanka tax collection settings for their applications.

## Action Items

- Review Stripe Tax API documentation for Sri Lanka configurations
- Update tax collection logic for digital service offerings in Sri Lanka
- Configure appropriate tax registration and VAT number handling if serving Sri Lankan customers

---

### Serverless environment version 5 is now available
- Published: 2026-02-25T00:00:00+00:00
- Products: 
- Tags: release, feature, api, performance
- URL: https://standarddb.com/databricks/databricks-releases-serverless-environment-version-5-with-enhanced-sql-and--vzmhbmxy
- Original: https://docs.databricks.com/aws/en/release-notes/serverless

## Serverless Environment Version 5 Release

Databricks has released serverless environment version 5, making it available immediately for all users running serverless compute across notebooks and jobs. This update is available in both CPU and GPU environment variants, with automatic rollout ensuring all users receive the upgrade over a short period.

### Key New Features

**SQL Enhancements:**
- **LIMIT ALL support for recursive CTEs**: You can now use the `LIMIT ALL` clause with recursive common table expressions to explicitly specify that no row limit should be applied to query results.
- **EXECUTE IMMEDIATE with constant expressions**: The `EXECUTE IMMEDIATE` statement now supports constant expressions in the query string, enabling more flexible dynamic SQL execution.
- **Improved error handling**: Attempting to append to existing files in Unity Catalog volumes now returns more descriptive error messages.

**Geospatial Functions:**
- **`st_dump` function**: Decompose geometry objects into their constituent parts, returning a set of simpler geometries.
- **Polygon interior ring functions**: Two new functions (`st_numinteriorrings` and `st_interiorringn`) for working with polygon interior rings.

### Configuration and Behavior Improvements

Users can now configure the `spark.sql.files.maxPartitionBytes` Spark configuration parameter on serverless compute to control partition byte packing when reading files. The release also includes behavior changes for improved metadata handling in materialized views and streaming tables, along with corrections for null struct handling in Parquet and schema operations.

### Automatic Deployment

As with all serverless compute releases, users automatically receive the latest version without requiring code changes. The stable client API based on Spark Connect ensures application compatibility across upgrades.

---

### WAF - WAF Release - Scheduled changes for 2026-03-02
- Published: 2026-02-25T00:00:00+00:00
- Products: Cloudflare
- Tags: release, security, bugfix, api
- URL: https://standarddb.com/cloudflare/cloudflare-waf-adds-three-new-detection-rules-for-smartermail-flaws-and-command--mnsgsdhh
- Original: https://developers.cloudflare.com/changelog/post/scheduled-waf-release

## New WAF Rule Releases

Cloudflare's Web Application Firewall is adding protection against emerging vulnerabilities and attack vectors in a scheduled release on **March 2, 2026**.

### New Detections

Two critical SmarterMail vulnerabilities are being addressed with new WAF rules:

- **SmarterMail - Arbitrary File Upload (CVE-2025-52691)**: New rule to detect attempts to exploit file upload vulnerabilities in SmarterMail
- **SmarterMail - Authentication Bypass (CVE-2026-23760)**: New rule to detect authentication bypass attacks against SmarterMail

Additionally, a new **Command Injection - Nslookup (Beta)** rule will be released. This beta rule will eventually be merged into the existing "Command Injection - Nslookup" rule.

### Rollout Behavior

All new rules will deploy in **Log mode**, meaning they will identify and log matching traffic without blocking it. This allows customers to evaluate the rules and adjust configurations before enforcement. Customers can manually adjust rule sensitivity settings based on their specific security posture and application requirements.

---

### Activity log now tracks 100% of team and project changes
- Published: 2026-02-25T00:00:00+00:00
- Products: Vercel
- Tags: feature, platform
- URL: https://standarddb.com/vercel/vercel-activity-log-now-tracks-100-of-team-and-project-changes-with-88-new--yrsgkgtd
- Original: https://vercel.com/changelog/activity-log-now-tracks-100-of-team-and-project-changes

## Enhanced Activity Log Coverage

Vercel's activity log has been significantly expanded to provide **complete visibility** into all team and project changes. Previously, some settings modifications were untracked, creating potential gaps in audit trails and making it difficult to maintain comprehensive change history.

## What Changed

With the addition of **88 new events**, the activity log now captures **100% coverage** of all actions performed on your Vercel team and projects. This includes:

- Team settings modifications
- Project configuration changes
- Permission and access updates
- All other previously untracked administrative actions

## For Developers and Teams

This enhancement is particularly valuable for:

- **Compliance and auditing**: Full trail of who made what changes and when
- **Security monitoring**: Track unauthorized or unexpected configuration changes
- **Team accountability**: Understand the complete history of project modifications
- **Troubleshooting**: Correlate issues with configuration changes

You can access the activity log directly in your Vercel dashboard. Learn more in the [official documentation](https://vercel.com/docs/activity-log).

---

### Containers - Run 15x more Containers with higher resource limits
- Published: 2026-02-25T00:00:00+00:00
- Products: Cloudflare
- Tags: platform, feature
- URL: https://standarddb.com/cloudflare/cloudflare-containers-now-support-15x-higher-resource-limits-up-to-6tb-memory--kdrdqtbb
- Original: https://developers.cloudflare.com/changelog/post/2026-02-25-higher-container-resource-limits

## Expanded Resource Limits for Container Workloads

Cloudflare Containers now support dramatically increased resource allocations across memory, vCPU, and disk capacity. This represents a 15x expansion in overall concurrent capacity, enabling developers to run significantly larger and more resource-intensive containerized applications on the Cloudflare edge.

### New Limits

The following resource limits have been updated:

- **Memory**: Increased from 400GB to 6TB per concurrent live instance pool
- **vCPU**: Increased from 100 to 1,500 per concurrent live instance pool  
- **Disk**: Increased from 2TB to 30TB per concurrent live instance pool

### What This Enables

With these expanded limits, developers can now run:

- Up to 15,000 instances of the `lite` instance type concurrently
- Up to 6,000 instances of `basic` type
- Over 1,500 instances of `standard-1` type
- Over 1,000 instances of `standard-2` type

This makes Containers practical for larger-scale workloads that previously required external container orchestration solutions.

### Next Steps

Review the [Containers Limits documentation](https://developers.cloudflare.com/containers/platform-details/limits/) for detailed information on all available instance types and their resource specifications.

---

### Agents, Workers - Agents SDK v0.6.0: RPC transport for MCP, optional OAuth, hardened schema conversion, and @cloudflare/ai-chat fixes
- Published: 2026-02-25T00:00:00+00:00
- Products: Cloudflare Workers
- Tags: release, feature, api, sdk, performance
- URL: https://standarddb.com/cloudflare/agents-sdk-v0-6-0-adds-rpc-transport-for-mcp-optional-oauth-and-schema-hardening--tvzzzzqm
- Original: https://developers.cloudflare.com/changelog/post/2026-02-25-agents-sdk-v060

## RPC Transport for MCP Connections

The Agents SDK now supports connecting an Agent to an McpAgent within the same Worker using Durable Object bindings instead of HTTP URLs. This eliminates network overhead and keeps all communication within the Cloudflare runtime.

Key features of the RPC transport:
- **Zero-network connections** — Pass the Durable Object namespace directly to `addMcpServer()` without HTTP serialization
- **Hibernation support** — RPC connections automatically survive Durable Object hibernation with persistent binding names and context props
- **Deduplication** — Calling `addMcpServer()` with the same server name returns the existing connection, with stable IDs across hibernation
- **Smaller footprint** — Transport internals reduced from 609 lines to 245 lines, using MCP SDK's `JSONRPCMessageSchema` for validation

## Simplified OAuth Configuration

OAuth is now opt-in rather than automatic. Servers that don't require authentication can be connected with a single line of code:

```javascript
await this.addMcpServer("my-server", "https://mcp.example.com");
```

When a server requires authentication, the SDK provides a clear error message guiding developers to add OAuth config. The restore-from-storage flow gracefully skips auth provider creation for non-OAuth servers.

## Production-Ready Schema Conversion

The JSON Schema to TypeScript converter has been hardened to handle edge cases that previously caused crashes:

- Circular reference and depth guards prevent stack overflows
- `$ref` resolution supports JSON Pointers (`#/definitions/...`, `#/$defs/...`)
- Support for JSON Schema 2020-12 `prefixItems` and OpenAPI 3.0 `nullable: true`
- Per-tool error isolation prevents one malformed schema from crashing the full pipeline
- Fallback to `{ type: "object" }` when `inputSchema` is missing

## `@cloudflare/ai-chat` Reliability Improvements

A batch of bug fixes improve streaming and message handling:

- **Tool denial flow** — Denied tool approvals now properly transition to `output-denied` state for Anthropic provider compatibility
- **Abort/cancel support** — Streaming responses correctly cancel the reader loop and signal completion to clients
- **Message deduplication** — `persistMessages()` reconciles duplicate assistant messages by content and order
- **Error tagging** — `requestId` in `OnChatMessageOptions` enables properly-tagged error responses
- **Thinking preservation** — Anthropic `redacted_thinking` blocks are no longer stripped by the sanitizer
- **Endpoint reliability** — `/get-messages` now works correctly when users override `onRequest()` without calling `super.onRequest()`
- **Undeprecated client APIs** — `createToolsFromClientSchemas`, `clientTools`, and related APIs restored for runtime-defined tools
- **Schema initialization** — Fixed errors when calling `getAITools()` in `onChatMessage`

## Getting Started

Update to the latest version with:

```bash
npm i agents@latest @cloudflare/ai-chat@latest
```

---

### Workers - No config? No problem. Just `wrangler deploy`
- Published: 2026-02-25T00:00:00+00:00
- Products: Cloudflare Workers
- Tags: feature, release, sdk
- URL: https://standarddb.com/cloudflare/wrangler-4-68-0-brings-automatic-configuration-projects-deploy-without-config--ywgdxtrt
- Original: https://developers.cloudflare.com/changelog/post/2026-02-25-wrangler-autoconfig-ga

## Automatic Configuration Now Generally Available

Starting with Wrangler 4.68.0, you can deploy any project to Cloudflare Workers with a single command—`wrangler deploy`—without needing to maintain a configuration file. The CLI now automatically detects your framework, installs required adapters, and sets everything up for you.

## How It Works

When you run `wrangler deploy` in a project directory without a configuration file, Wrangler performs the following steps:

- **Detects your framework** from `package.json`
- **Prompts you to confirm** the detected settings
- **Installs required adapters** automatically
- **Generates a `wrangler.jsonc`** configuration file
- **Deploys your project** to Cloudflare Workers

You can also use `wrangler setup` to configure without deploying, or pass the `--yes` flag to skip interactive prompts.

## Dashboard Integration

When you connect a repository through the Cloudflare dashboard, the system automatically generates a pull request with all necessary files and includes a preview deployment so you can review changes before merging.

## Timeline

This feature was introduced as an experimental capability in December 2025 and is now the default behavior for all users. The graduation from experimental to general availability removes friction from the Workers deployment experience and makes it more accessible to developers new to the platform.

---

### Workers - Write structured queries to filter and search your Workers logs and traces
- Published: 2026-02-25T00:00:00+00:00
- Products: Cloudflare Workers
- Tags: feature, platform
- URL: https://standarddb.com/cloudflare/cloudflare-workers-observability-adds-structured-query-language-for-logs-and--jvhdzysc
- Original: https://developers.cloudflare.com/changelog/post/2026-02-24-observability-query-language

## Query Language for Workers Observability

Cloudflare has added a query language to Workers Observability, enabling developers to write structured queries directly in the search bar to filter logs and traces. The search bar functions as both a free text search box for broad keyword searches and a field-level query interface for precise filtering.

## Key Features

The query language supports:

- **Free text search** — search everywhere with keywords like `error`, or match exact phrases with `"exact phrase"`
- **Field queries** — filter by specific fields using comparison operators (e.g., `status = 500` or `$workers.wallTimeMs > 100`)
- **Operators** — `=`, `!=`, `>`, `>=`, `<`, `<=`, and `:` (contains)
- **Functions** — `contains(field, value)`, `startsWith(field, prefix)`, `regex(field, pattern)`, and `exists(field)`
- **Boolean logic** — combine conditions with `AND`, `OR`, and `NOT`

## Developer Workflow

Queries written in the search bar sync with the Query Builder sidebar, allowing developers to write queries by hand and refine them visually, or build filters in the Query Builder and see the corresponding query syntax. The search bar provides autocomplete suggestions for metadata fields and operators as you type, reducing the learning curve for new users.

Access the query language via the [Workers Observability dashboard](https://dash.cloudflare.com/?to=/:account/workers-and-pages/observability/), where a help icon provides a full syntax reference including all supported operators, functions, and keyboard shortcuts.

---

### Radar - RPKI ASPA Deployment Insights on Cloudflare Radar
- Published: 2026-02-25T00:00:00+00:00
- Products: Cloudflare
- Tags: feature, api, platform
- URL: https://standarddb.com/cloudflare/cloudflare-radar-adds-rpki-aspa-deployment-tracking-with-new-api-endpoints--jkgrtqzk
- Original: https://developers.cloudflare.com/changelog/post/2026-02-25-radar-aspa-insights

## New ASPA Capabilities in Cloudflare Radar

Cloudflare Radar now tracks **Autonomous System Provider Authorization (ASPA)** deployment insights, offering network operators unprecedented visibility into ASPA object adoption and verification across the internet's routing infrastructure.

### New API Endpoints

Developers and network operators can now access ASPA data programmatically through three new API endpoints:

- **`/bgp/rpki/aspa/snapshot`** — Retrieve current or historical ASPA objects
- **`/bgp/rpki/aspa/changes`** — Track modifications to ASPA objects over time
- **`/bgp/rpki/aspa/timeseries`** — Access ASPA object counts as time-series data

These endpoints integrate into Radar's existing BGP and RPKI infrastructure, allowing automation and integration into existing monitoring systems.

### Enhanced Dashboard Visualizations

The global routing page now displays:

- **ASPA Deployment Trend** — A daily-granularity chart showing cumulative ASPA object adoption over time
- **ASPA Objects Table** — Searchable list of active ASPA objects by ASN or AS name

Country and region-specific routing pages include a new **ASPA Deployment Rate widget** showing the adoption percentage for ASNs registered in that geography.

On individual AS routing pages, the connectivity table now marks **ASPA-verified upstreams** with checkmarks, includes a dedicated ASPA upstreams table, and displays a daily-granularity timeline of ASPA changes.

### Next Steps

Network operators can explore the new ASPA data on [Radar's routing page](https://radar.cloudflare.com/routing) or integrate the new API endpoints into operational dashboards for monitoring BGP security posture across their networks.

---

### February 25, 2026
- Published: 2026-02-25T00:00:00+00:00
- Products: 
- Tags: feature, api, platform
- URL: https://standarddb.com/deepgram/deepgram-voice-agent-adds-updatethink-for-mid-conversation-llm-provider--ywmklyky
- Original: https://developers.deepgram.com/changelog/2026/2/25

## Dynamic Provider Switching

Deepgram's Voice Agent now supports the `UpdateThink` client message, enabling developers to completely reconfigure the Think provider during active conversations. This complements the existing `UpdatePrompt` message by allowing full configuration replacement rather than just prompt appending.

## Key Capabilities

With `UpdateThink`, you can dynamically:
- **Switch LLM providers** – Change between different AI model providers mid-conversation
- **Update the model** – Select a different model within a provider (e.g., gpt-4o-mini)
- **Replace the prompt** – Set an entirely new system prompt for the agent
- **Reconfigure functions** – Modify the available functions and tools available to the agent

All these changes occur within a single message, providing seamless control over the agent's reasoning setup.

## Implementation

The message format is straightforward, with developers specifying the new Think configuration:

```json
{
    "type": "UpdateThink",
    "think": {
        "provider": {
            "type": "open_ai",
            "model": "gpt-4o-mini"
        },
        "prompt": "You are a helpful voice assistant."
    }
}
```

The Deepgram server confirms successful updates with a `ThinkUpdated` response message. This allows developers to implement adaptive conversation flows, A/B test different reasoning approaches, or adjust agent behavior based on conversation context in real-time.

For implementation details and advanced configuration options, refer to the [UpdateThink documentation](https://developers.deepgram.com/docs/voice-agent-update-think).

---

### A New Era for Sourcegraph: The Intelligence Layer for AI Coding Agents and Developers
- Published: 2026-02-24T23:20:01+00:00
- Products: 
- Tags: release, feature, api, platform, integration
- URL: https://standarddb.com/sourcegraph/sourcegraph-7-0-debuts-mcp-integration-for-ai-agents-adds-semantic-search--ctggmmct
- Original: https://webflow.sourcegraph.com/blog/a-new-era-for-sourcegraph-the-intelligence-layer-for-ai-coding-agents-and-developers

## Sourcegraph Pivots Toward AI Agent Enablement

Sourcegraph 7.0 marks a strategic shift for the company, moving from a developer-focused code search tool to a foundational intelligence layer designed for both human developers and AI coding agents. The release addresses a critical gap in current agentic workflows: agents often lack the deep, contextual understanding of large enterprise codebases that they need to operate reliably.

## MCP Integration and Deep Search Capabilities

The headline feature is **Deep Search via the Model Context Protocol (MCP)**, which allows AI agents to query enterprise codebases with semantic, cross-repository, historical, and architectural context. Key capabilities include:

- **Cross-repository reasoning**: Understand how systems connect across repository boundaries
- **Architectural navigation**: Reason about why code was written and navigate complex patterns with confidence
- **Historical context**: Access the reasoning behind past architectural decisions
- **Shared semantics**: Both developers and agents access the same underlying code intelligence

The company has also improved its existing Deep Search functionality for more accurate results and broader workflow integration.

## Foundation and Performance Improvements

Beyond the headline capabilities, Sourcegraph 7.0 includes several foundational updates:

- Modernized Svelte UI for improved responsiveness
- Streamlined instance setup flow for faster team onboarding
- Weekly shipping cadence for Sourcegraph Cloud customers
- Refined product focus centered on reliable code intelligence

## Positioning and Honest Limitations

Sourcegraph explicitly sets realistic expectations: the release is not claiming agents produce perfect code, that Sourcegraph replaces human judgment, or offering a major UX redesign. Instead, the focus is on **infrastructure correctness**—providing reliable, deep context that enables both developers and agents to work effectively with production code at scale.

For existing customers, 7.0 unlocks new value from already-indexed codebases through AI agent support. For organizations evaluating enterprise-scale agent workflows, the platform becomes authoritative shared infrastructure.

---

### CodeQL adds Go 1.26 and Kotlin 2.3.10 support and improves query accuracy
- Published: 2026-02-24T22:20:46+00:00
- Products: GitHub
- Tags: release, feature, api, security
- URL: https://standarddb.com/github/codeql-2-24-2-adds-go-1-26-and-kotlin-2-3-10-support-improves-query-accuracy--lgzkpwdz
- Original: https://github.blog/changelog/2026-02-24-codeql-adds-go-1-26-and-kotlin-2-3-10-support-and-improves-query-accuracy

## Language Support Expansion

CodeQL 2.24.2 extends its static analysis capabilities to the latest versions of two major programming languages. **Go 1.26** is now fully supported for security analysis, enabling developers using the latest Go release to benefit from GitHub code scanning. Similarly, **Kotlin support has been updated to version 2.3.10**, ensuring comprehensive coverage for modern Kotlin development.

## Query Accuracy Improvements

The release includes targeted refinements to reduce false positives and improve detection accuracy:

**C# CSRF Detection**: The `cs/web/missing-token-validation` query now properly recognizes antiforgery attributes applied to base controller classes. This fix addresses cases where `[ValidateAntiForgeryToken]` or `[AutoValidateAntiforgeryToken]` decorators on parent classes were previously missed, reducing false positives in ASP.NET applications.

**Java/Kotlin Pattern Validation**: Queries analyzing injection vulnerabilities (`java/ssrf`, `java/path-injection`, `java/log-injection`) now recognize `@javax.validation.constraints.Pattern` annotations as sanitizers. This enhancement allows developers to use standard validation annotations more effectively without triggering unnecessary security warnings.

## Azure SDK and Availability

Python analysis now includes request forgery sink models for the Azure SDK, improving detection of potential SSRF vulnerabilities in cloud-based Python applications. The new functionality is automatically deployed to GitHub.com users and will be included in future GitHub Enterprise Server releases. Administrators running older GHES versions can manually upgrade their CodeQL installation.



---

### Dependabot can group updates by dependency name across multiple directories
- Published: 2026-02-24T21:28:17+00:00
- Products: GitHub Dependabot
- Tags: feature, platform
- URL: https://standarddb.com/github/dependabot-now-consolidates-cross-directory-dependency-updates-into-single-pull--ljkyslsy
- Original: https://github.blog/changelog/2026-02-24-dependabot-can-group-updates-by-dependency-name-across-multiple-directories

## Cross-Directory Dependency Grouping

Dependabot has added the ability to group updates by dependency name across multiple directories, streamlining dependency management in repositories with complex structures. Previously, if a dependency needed updating in multiple directories, Dependabot would generate a separate pull request for each directory.

## Before and After

**Previous behavior:** Upgrading a single dependency like `requests` across three directories would create three individual pull requests:
- `chore(deps): bump requests in /service-a`
- `chore(deps): bump requests in /service-b`
- `chore(deps): bump requests in /service-c`

**New behavior:** These updates are now consolidated into a single pull request, regardless of the number of affected directories.

## Impact and Use Cases

This change is **especially valuable for monorepos** where a single dependency bump can touch dozens of directories. The consolidation reduces noise in pull request queues and simplifies the review and merge workflow.

## Getting Started

Configure directory groups in your repository's `dependabot.yml` file using the updated [Dependabot configuration documentation](https://docs.github.com/code-security/dependabot/dependabot-version-updates/configuration-options-for-the-dependabot.yml-file#groups). This feature is available for all github.com users and will ship in GitHub Enterprise Server (GHES) 3.21.

---

### How we rebuilt Next.js with AI in one week
- Published: 2026-02-24T20:00:00+00:00
- Products: Cloudflare Workers
- Tags: release, feature, platform, integration, performance
- URL: https://standarddb.com/cloudflare/cloudflare-releases-vinext-next-js-compatible-framework-built-on-vite-that--qkbvgqpx
- Original: https://blog.cloudflare.com/vinext

## Vinext: A Clean Reimplementation of Next.js on Vite

Cloudflare has released **Vinext**, a new framework that reimplements the Next.js API surface on top of Vite rather than wrapping or adapting Next.js output. Unlike adapters like OpenNext that reverse-engineer Next.js builds, Vinext is a ground-up alternative implementation supporting routing, server rendering, React Server Components, server actions, caching, and middleware—all built as a Vite plugin.

The framework works as a drop-in replacement: replace `next` with `vinext` in your npm scripts and existing `app/`, `pages/`, and `next.config.js` files work unchanged. Three commands handle development, production builds, and deployment to Cloudflare Workers.

## Performance Benchmarks

Using a 33-route App Router test application, Vinext demonstrates substantial performance gains:

**Production Build Times:**
- Next.js 16.1.6: 7.38s (baseline)
- Vinext with Rollup: 4.64s (1.6x faster)
- Vinext with Rolldown: 1.67s (4.4x faster)

**Client Bundle Size (gzipped):**
- Next.js 16.1.6: 168.9 KB (baseline)
- Vinext: 72.9 KB (57% smaller)

Cloudflare notes these are directional benchmarks from a single test fixture and should evolve as projects develop. Full methodology and historical results are published publicly.

## Cloudflare Workers as First-Class Deployment Target

Vinext is purpose-built for Cloudflare Workers, offering seamless deployment with a single command: `vinext deploy`. Both App Router and Pages Router are supported with full client-side hydration, interactive components, and client-side navigation.

For caching, Vinext includes a Cloudflare KV cache handler providing Incremental Static Regeneration (ISR) out of the box. The caching layer is pluggable—developers can swap in R2, Cache API, or custom backends depending on their application needs.

## Native Platform Integration

A key advantage over next dev is that development now runs on Cloudflare's workerd runtime, eliminating the need for workarounds like `getPlatformProxy`. This allows developers to test platform-specific APIs—Durable Objects, KV, AI bindings—directly during development without Node.js-specific limitations.

The project was built in one week by a single engineer with AI assistance, costing approximately $1,100 in tokens. Several examples are already running in production, including an App Router playground, Hacker News clone, and a demonstration of Cloudflare Agents running in a Next.js app.

---

### Replit Pro Is Here — and Core Now Offers Even Better Value
- Published: 2026-02-24T20:00:00+00:00
- Products: 
- Tags: pricing, feature, release
- URL: https://standarddb.com/replit/replit-launches-pro-tier-at-100-month-core-drops-to-20-month-with-team--dvzddvfm
- Original: https://blog.replit.com/pro-plan

## New Pricing Structure

Replit has announced a restructured pricing strategy with two main paid tiers replacing its existing plan lineup. **Replit Pro** ($100/month) is positioned as the premium offering for production-grade developers, while **Replit Core** ($20/month, down from $25) becomes the entry-level paid plan focused on getting started and casual building.

## Pro Plan Benefits

The Pro tier includes significant upgrades for serious builders:

- **Turbo Mode** for Agent with up to 2x faster performance and access to the most capable models
- **Tiered credit packages** ranging from $100 to $4,000/month with volume discounts and monthly rollover
- **Team scaling** supporting up to 15 builders with pooled credits and no per-seat fees
- **Extended data retention** of 28 days (vs. 7 days on other plans)
- **Priority support** with 24-hour response times
- **Exclusive community** access for sharing and learning

## Core Plan Improvements

The Core plan price cut is accompanied by new features to improve accessibility:

- **Team collaboration** now included, allowing up to 5 collaborators per workspace (previously required a separate Teams subscription)
- **Cost control options** including Economy Mode (roughly one-third the cost of Agent 3) and Power Mode for performance-optimized tasks
- Automatic pricing reduction for existing subscribers at next renewal

## Agent Performance Modes

All paid plans now have access to new Agent modes to balance cost and capability:

- **Economy Mode:** Cost-optimized, approximately 33% of previous pricing
- **Power Mode:** Performance-optimized for complex tasks and larger codebases
- **Turbo Mode:** 2x faster than Power Mode, available only on Pro and Enterprise (costs up to 6x more)

## Team Migration and Collaboration Updates

The existing Teams plan is being sunset, with current subscribers automatically upgraded to Pro at no additional cost through their current term. Monthly subscribers will begin paying the $100/month rate at next renewal, while yearly subscribers receive $100 in monthly credits through their term end.

All paid plans now feature improved collaboration infrastructure including personal and collaborative workspaces with unified billing, usage tracking per workspace, and simplified sharing controls.

These changes roll out to Core customers immediately and to Teams/Enterprise customers starting March 3, 2026.

---

### GitHub Code Quality: Organization-level dashboard in public preview
- Published: 2026-02-24T19:49:29+00:00
- Products: GitHub
- Tags: feature, platform
- URL: https://standarddb.com/github/github-code-quality-adds-organization-level-dashboard-for-cross-repository--hbdmyrjl
- Original: https://github.blog/changelog/2026-02-24-github-code-quality-organization-level-dashboard-in-public-preview

## Organization-Wide Code Quality Visibility

GitHub Code Quality has expanded with a new organization-level dashboard available in public preview. This feature enables organization owners, administrators, and developers to track and compare code health across all repositories where code quality analysis is enabled, providing a unified view of maintainability and reliability metrics.

## Key Features

The dashboard provides several critical capabilities:

- **Cross-repository comparison**: View a bubble chart showing code quality scores plotted by maintainability versus reliability across repositories
- **Filtering and sorting**: Organize repositories by quality attributes to quickly identify high-risk or low-quality codebases
- **Permission-aware access**: Organization owners and administrators see all repositories, while developers see only repositories where they have write access
- **Quick assessment**: Identify which repositories need attention and prioritize code quality improvements

## Access and Availability

To access the dashboard, navigate to the **Security** tab at the organization level and select **Code quality**. The feature is currently available to:

- Organizations in the GitHub Code Quality public preview on github.com
- GitHub Enterprise Cloud with data residency

For detailed information, refer to the [GitHub Code Quality documentation](https://docs.github.com/code-security/code-quality). Organizations can join the discussion and provide feedback in the [GitHub Code Quality announcement](https://github.com/orgs/community/discussions/177488).

---

### Repository dashboard is now generally available
- Published: 2026-02-24T18:49:44+00:00
- Products: GitHub
- Tags: release, feature, platform
- URL: https://standarddb.com/github/github-launches-repository-dashboard-for-general-availability-with-admin-access--fdmybvnm
- Original: https://github.blog/changelog/2026-02-24-repository-dashboard-is-now-generally-available

## Repository Dashboard Now Generally Available

GitHub has released its **repository dashboard** as a generally available feature, graduating from public preview. The dashboard simplifies repository discovery and management by providing a centralized hub for finding, filtering, and organizing repositories across personal projects and enterprise codebases.

## Key Additions in GA Release

This general availability release introduces two significant enhancements:

- **Admin Access View:** A new built-in view automatically surfaces all repositories where you have admin permissions, streamlining access control management and configuration across multiple projects.
- **Command Palette Integration:** The dashboard is now directly accessible through GitHub's command palette, allowing faster navigation without interrupting your current workflow.

## Core Features and Capabilities

The dashboard includes several powerful features for repository management:

- **Unified Repository Discovery:** Access any repository you've contributed to or have access to in a single, searchable location.
- **Curated Default Views:** Start with pre-built views including `My contributions`, `My repositories`, `My forks`, and the new `Admin access` view.
- **Advanced Filtering:** Refine searches using filters like visibility, language, organization, and more. Sort by relevance to identify your most active repositories.
- **Saved Custom Queries:** Bookmark frequently-used filter combinations to instantly return to your most important repository views.

## Access Methods

You can access the repository dashboard through multiple routes:
- Visit [github.com/repos](https://github.com/repos) directly
- Click the repositories icon in the top global navigation bar
- Select **Repositories** from the left-hand navigation menu
- Search for "Repositories" in the command palette

---

### Slack Agent Skill simplifies building Slack agents with coding assistants
- Published: 2026-02-24T13:00:00+00:00
- Products: Vercel
- Tags: feature, integration, platform
- URL: https://standarddb.com/vercel/vercel-launches-slack-agent-skill-to-simplify-agent-deployment-with-one-click--npshgfty
- Original: https://vercel.com/changelog/slack-agent-skill-simplifies-building-slack-agents-with-coding-assistants

## Slack Agent Skill Now Available

Vercel has released the **Slack Agent Skill**, a new tool that streamlines the process of building and deploying Slack agents through coding assistants. The skill handles all the infrastructure complexity typically required for Slack integrations, enabling developers to focus entirely on agent behavior and logic.

### What's Included

The skill automates the entire setup workflow across five key stages:

1. **Project Setup** — Select your LLM provider and initialize from the Slack Agent Template
2. **Slack App Creation** — Auto-generate a customized app manifest and register it in Slack's console
3. **Environment Configuration** — Securely configure signing secrets, bot tokens, and API keys with built-in validation
4. **Local Testing** — Test locally using ngrok integration and verify the Slack connection
5. **Production Deployment** — Deploy to Vercel with all environment variables automatically configured

### Getting Started

Install the skill using npm and invoke it directly in your coding agent:

```bash
npx skills add vercel-labs/slack-agent-skill
```

Then invoke the wizard in your agent (for example, `/slack-agent new` in Claude Code) to step through the guided setup process.

### Next Steps

Developers can either use the provided [Slack Agent Template](https://vercel.com/templates/nitro/slack-agent-template) to deploy immediately and customize later, or build a custom agent from scratch. Free tier deployment is available on Vercel, with Pro and Enterprise options for production workloads.

---

### GPT 5.3 Codex is now on AI Gateway
- Published: 2026-02-24T13:00:00+00:00
- Products: Vercel AI SDK, Vercel
- Tags: model, integration, feature, api
- URL: https://standarddb.com/vercel/gpt-5-3-codex-arrives-on-vercel-ai-gateway-with-25-faster-execution--ycbppgrb
- Original: https://vercel.com/changelog/gpt-5-3-codex-is-now-on-ai-gateway

## GPT 5.3 Codex Now Available

Vercel has introduced GPT 5.3 Codex to its AI Gateway platform, a new model designed specifically for developers building AI-powered applications. The model merges the coding expertise of GPT-5.2-Codex with the reasoning depth of GPT-5.2, delivering a more capable tool for complex software development tasks.

## Key Improvements

**Performance and Efficiency:**
- 25% faster execution than its predecessors
- Improved token efficiency for cost-effective operations
- Built for long-running agentic work requiring research, tool use, and multi-step execution

**Developer Experience:**
- Better understanding of underspecified prompts in web development contexts
- Defaults to production-ready output without requiring excessive refinement
- Mid-task steering capability without losing context, enabling iterative development workflows

## How to Use

To integrate GPT 5.3 Codex into your applications, set the model identifier to `openai/gpt-5.3-codex` in the Vercel AI SDK:

```tsx
import { streamText } from 'ai';

const result = streamText({
  model: 'openai/gpt-5.3-codex',
  prompt: 'Research our current API architecture, identify performance bottlenecks, refactor the slow endpoints, add monitoring, and deploy the changes to staging.',
});
```

## AI Gateway Features

The model integrates with AI Gateway's existing capabilities, including built-in observability, Bring Your Own Key support, intelligent provider routing, and automatic retries for improved reliability and uptime beyond individual provider guarantees.

---

### Python Vercel Functions bundle size limit increased to 500MB
- Published: 2026-02-24T13:00:00+00:00
- Products: Vercel
- Tags: feature, platform
- URL: https://standarddb.com/vercel/vercel-doubles-python-functions-bundle-size-limit-to-500mb--wqgyxqbf
- Original: https://vercel.com/changelog/python-vercel-functions-bundle-size-limit-increased-to-500mb

## Python Functions Bundle Size Increase

Vercel has announced a significant increase to the bundle size limits for their Python runtime. The maximum uncompressed deployment bundle size for Vercel Functions using Python has been doubled from **250MB to 500MB**.

### What's Changed

- **Previous limit**: 250MB uncompressed bundle size
- **New limit**: 500MB uncompressed bundle size
- **Runtime**: Applies to Python-based Vercel Functions

### Impact for Developers

This increase is particularly beneficial for Python developers who rely on:
- **Large ML/data science libraries** (NumPy, Pandas, scikit-learn, etc.)
- **Complex dependency trees** with many transitive dependencies
- **Compiled extensions** that require substantial disk space
- **Multi-service applications** deployed as single functions

The doubled limit removes a significant constraint for Python deployments on Vercel, making it viable to deploy more feature-rich applications without having to split them across multiple functions or externalize large dependencies.

### Getting Started

Developers can begin deploying Python applications immediately using the new limits. Vercel provides templates for popular Python frameworks:
- [FastAPI Python Boilerplate](https://vercel.com/templates/python/fastapi-python-boilerplate)
- [Flask Python Boilerplate](https://vercel.com/templates/python/flask-python-boilerplate)

For more details, see the [Functions limitations documentation](https://vercel.com/docs/functions/limitations).

---

### Zero Trust WARP Client - WARP client for Windows (version 2026.1.150.0)
- Published: 2026-02-24T01:15:23+00:00
- Products: Cloudflare
- Tags: release, bugfix, feature, platform
- URL: https://standarddb.com/cloudflare/cloudflare-warp-client-for-windows-reaches-ga-with-multi-user-mode-improvements--bnpycmdn
- Original: https://developers.cloudflare.com/changelog/post/2026-02-24-warp-windows-ga

## Overview

A new general availability release for the Cloudflare WARP client on Windows is now available. Version 2026.1.150.0 includes multiple bug fixes, stability improvements, and new configuration options for enterprise deployments.

## Key Changes and Features

**Multi-User Mode Improvements**
- Fixed an issue where switching from pre-login registration to user registration could cause Mobile Device Management (MDM) configuration association to be lost

**NetBIOS Over TCP/IP Management**
- Added a new feature to manage NetBIOS over TCP/IP functionality on Windows clients
- NetBIOS over TCP/IP is now **disabled by default** and can be enabled through device profile settings
- This provides better control for security-conscious enterprise deployments

**Bug Fixes and Stability**
- Resolved local network exclusion feature failures when configured with a timeout of `0`
- Fixed DNS request failures affecting clients in Traffic and DNS mode
- Improved client certificate posture check logging to only report results after user login
- Fixed misconfigured DEX HTTP tests that were preventing new registrations
- Enhanced device shutdown behavior when the daemon becomes unresponsive
- Improved accuracy of device colocation information reporting in the Cloudflare One dashboard

## Known Issues

Users should be aware of three documented issues:
- Windows 11 24H2 users may experience performance issues (Microsoft KB5062553 or later recommended)
- A false positive antimalware warning may appear (update Microsoft Security Intelligence to 1.429.19.0+)
- DNS resolution may break in Secure Web Gateway tunnel-only mode with custom DNS servers (workaround: toggle WARP off and back on)

## Getting Started

The stable release is available on the [WARP downloads page](https://developers.cloudflare.com/cloudflare-one/team-and-resources/devices/warp/download-warp/). Enterprise customers using MDM deployment should review the multi-user mode documentation to understand the fixes.

---

### Zero Trust WARP Client - WARP client for macOS (version 2026.1.150.0)
- Published: 2026-02-24T01:15:22+00:00
- Products: Cloudflare
- Tags: release, bugfix, platform
- URL: https://standarddb.com/cloudflare/cloudflare-releases-warp-macos-client-v2026-1-150-0-with-dns-tunnel-fixes--mzkvnlfv
- Original: https://developers.cloudflare.com/changelog/post/2026-02-24-warp-macos-ga

## WARP macOS Client v2026.1.150.0 Now Available

Cloudflare has released version 2026.1.150.0 of the WARP client for macOS, a maintenance update addressing key stability issues. The release is available on the [stable releases downloads page](https://developers.cloudflare.com/cloudflare-one/team-and-resources/devices/warp/download-warp/).

### Bug Fixes & Improvements

This release resolves multiple connectivity and configuration issues:

- **Local network exclusion timeout fix**: Corrected a bug where the local network exclusion feature failed when configured with a timeout value of `0`
- **Device colocation reporting**: Enhanced accuracy of device colocation information displayed in the Cloudflare One dashboard
- **DNS server configuration**: Fixed failures in DNS server configuration that previously caused tunnel connection delays
- **DEX HTTP tests**: Resolved an issue where misconfigured DEX HTTP tests blocked new device registrations
- **Traffic and DNS mode**: Fixed DNS request failures affecting clients operating in Traffic and DNS mode

### What You Need to Do

macOS users with WARP client deployments should update to this version to ensure proper functionality of local network exclusion policies and DNS configuration. The fixes are particularly important for organizations relying on DEX diagnostics and hybrid DNS/traffic modes in their Zero Trust architecture.

---

### v0.17.1-rc0: update mlx-c bindings to 0.5.0 (#14380)
- Published: 2026-02-24T00:44:29+00:00
- Products: 
- Tags: release, bugfix
- URL: https://standarddb.com/ollama/ollama-v0-17-1-rc0-updates-mlx-c-bindings-to-0-5-0--tcrycfwk
- Original: https://github.com/ollama/ollama/releases/tag/v0.17.1-rc0

## Ollama v0.17.1-rc0 Release

This release candidate includes a key update to the mlx-c bindings, advancing them from the previous version to **0.5.0**. The mlx-c bindings are crucial components that enable Ollama to interface with Apple's MLX framework on compatible hardware.

### Key Changes

- **mlx-c bindings updated to 0.5.0**: Improves compatibility and support for MLX-based inference
- **Linux build toolchain update**: Switches from GCC to GCC 11, ensuring better standards compliance and performance optimization on Linux systems

### What This Means for Users

Users running Ollama on Linux systems will benefit from improved compiler support and potential performance optimizations provided by GCC 11. The mlx-c binding update ensures better compatibility with Apple silicon systems and may include bug fixes or feature improvements from the upstream MLX library.

### Availability

This is a release candidate (rc0), meaning it's available for testing before the final v0.17.1 release. Users can test this build to help identify any potential issues before the stable release.

---

### Zero Trust WARP Client - WARP client for Linux (version 2026.1.150.0)
- Published: 2026-02-24T00:14:20+00:00
- Products: Cloudflare
- Tags: release, bugfix
- URL: https://standarddb.com/cloudflare/cloudflare-releases-warp-client-v2026-1-150-0-for-linux-with-dns-and-local--xwypsvlh
- Original: https://developers.cloudflare.com/changelog/post/2026-02-24-warp-linux-ga

## WARP Client v2026.1.150.0 Now Available for Linux

Cloudflare has released version 2026.1.150.0 of the Zero Trust WARP client for Linux, bringing several bug fixes and improvements for enterprise users. The release is now available on the stable releases downloads page.

### Key Fixes and Improvements

This maintenance release addresses several issues affecting production deployments:

- **DNS Reliability**: Fixed DNS request failures affecting clients running in Traffic and DNS mode or DNS-only mode
- **Local Network Exclusion**: Resolved an issue where the local network exclusion feature failed when configured with a timeout value of `0`
- **Device Reporting**: Enhanced accuracy of device colocation information displayed in the Cloudflare One dashboard
- **DEX HTTP Tests**: Fixed misconfiguration issues that prevented new device registrations

### Important Security Update Required

**Action required for existing Linux users**: The WARP client version 2025.8.779.0 introduced an updated public key for Linux packages in September 2025. Organizations that installed the client before September 12, 2025 must update their public key to ensure the repository remains functional after December 4, 2025. Instructions for updating the key are available at [pkg.cloudflareclient.com](https://pkg.cloudflareclient.com/).

This is a standard maintenance release suitable for production environments running the WARP client on Linux systems.

---

### Containers - Run 15x more Containers with higher resource limits
- Published: 2026-02-24T00:00:00+00:00
- Products: Cloudflare
- Tags: release, feature, platform
- URL: https://standarddb.com/cloudflare/cloudflare-containers-raises-resource-limits-15x-now-supports-up-to-6tb-memory--hrwqtyjc
- Original: https://developers.cloudflare.com/changelog/post/2026-02-24-higher-container-resource-limits

## Higher Resource Limits for Cloudflare Containers

Cloudflare has substantially increased the resource limits for its Containers platform, enabling developers to run significantly larger and more complex workloads at scale.

### New Limits

The platform now supports:
- **Memory**: 6TiB per concurrent live container instances (up from 400GiB)
- **vCPU**: 1,500 (up from 100)
- **Disk**: 30TB (up from 2TB)

This represents a **15x increase** across all major resource dimensions.

### What This Enables

With these expanded limits, developers can now concurrently run:
- 15,000 instances of the `lite` instance type
- 6,000 instances of the `basic` type
- 1,500+ instances of `standard-1`
- 1,000+ instances of `standard-2`

### Next Steps

Review the [Limits documentation](https://developers.cloudflare.com/containers/platform-details/limits/) for detailed information on available instance types and how these new limits apply to your workloads.

---

### Durable Objects, Workers - deleteAll() now deletes Durable Object alarm
- Published: 2026-02-24T00:00:00+00:00
- Products: Cloudflare Workers
- Tags: feature, api, bugfix
- URL: https://standarddb.com/cloudflare/cloudflare-s-deleteall-now-clears-durable-object-alarms-reducing-api-calls--kjssnrwy
- Original: https://developers.cloudflare.com/changelog/post/2026-02-24-deleteall-deletes-alarms

## Simplified Durable Object Cleanup

Cloudflare has updated the `deleteAll()` method for Durable Objects to automatically clear alarm metadata in addition to user-stored data. Previously, developers needed to make two separate API calls to fully clean up a Durable Object's storage.

### What Changed

**Before:** Clearing all storage required two calls
```
await this.ctx.storage.deleteAlarm();
await this.ctx.storage.deleteAll();
```

**Now:** A single call handles everything
```
await this.ctx.storage.deleteAll();
```

### Impact and Scope

This change simplifies storage management by eliminating the need for sequential API calls. Alarm usage in Durable Objects stores metadata within the object's storage layer, which now gets cleaned up automatically. The improvement applies to both **KV-backed** and **SQLite-backed** Durable Objects.

### How to Use

The new behavior is automatically enabled for Workers with a compatibility date of **2026-02-24** or later. Existing applications are unaffected unless explicitly updated to the new compatibility date. Developers should review their cleanup routines and can consolidate dual calls into a single `deleteAll()` operation.

---

### Cache - Asynchronous stale-while-revalidate
- Published: 2026-02-24T00:00:00+00:00
- Products: Cloudflare
- Tags: feature, performance, api
- URL: https://standarddb.com/cloudflare/cloudflare-rolls-out-asynchronous-stale-while-revalidate-eliminating-cache--hmkmtszn
- Original: https://developers.cloudflare.com/changelog/post/2026-02-24-async-stale-while-revalidate

## Overview

Cloudflare's `stale-while-revalidate` Cache-Control directive is now fully asynchronous. Previously, revalidation would only trigger on the next visitor request after an asset expired. Now, revalidation begins automatically at expiry, allowing Cloudflare to serve stale content immediately while refreshing assets in the background.

## Key Benefits

**Lower latency**: The first visitor after cache expiry is no longer blocked waiting for an origin round-trip. Stale (but safe) content is served instantly while the fresh copy is fetched asynchronously.

**Fresher content**: Since revalidation happens automatically at expiry rather than waiting for a request, cached assets are refreshed sooner and visitors are more likely to receive up-to-date content.

## Important Considerations

**Origin traffic increase**: Your origin may experience elevated traffic as Cloudflare proactively revalidates assets at expiry.

**Cache status changes**: All revalidation cache statuses are now reported as `UPDATING` or `HIT` instead of `MISS` or `REVALIDATED`. Update any monitoring or analytics systems that depend on these statuses.

## Availability & Rollout

This change is live for all **Free, Pro, and Business** zones. Approximately 75% of Enterprise zones have been migrated, with the remaining zones rolling out throughout the quarter.

## Getting Started

Ensure your origin includes the `stale-while-revalidate` directive in the `Cache-Control` header. See the [Cache-Control documentation](https://developers.cloudflare.com/cache/concepts/cache-control/#revalidation) for implementation details.

---

### Durable Objects, Workers - deleteAll() now deletes Durable Object alarm
- Published: 2026-02-24T00:00:00+00:00
- Products: Cloudflare Workers
- Tags: feature, api, bugfix
- URL: https://standarddb.com/cloudflare/cloudflare-workers-deleteall-now-clears-durable-object-alarms-automatically--dhhmgpkb
- Original: https://developers.cloudflare.com/changelog/2026-02-24-deleteall-deletes-alarms

## Simplified Storage Cleanup for Durable Objects

The `deleteAll()` method for Cloudflare Durable Objects now automatically deletes alarms along with user-stored data, streamlining cleanup operations for developers.

### What Changed

Previously, `deleteAll()` only removed user data from a Durable Object's storage. Since alarm metadata is stored within the object's storage, developers had to make two separate API calls to completely clear an object:

```javascript
// Old approach - two calls required
await this.ctx.storage.deleteAlarm();
await this.ctx.storage.deleteAll();
```

With this update, a single call now handles everything:

```javascript
// New approach - one call clears all
await this.ctx.storage.deleteAll();
```

### Compatibility and Coverage

This change applies to all Workers with a compatibility date of **2026-02-24 or later** and works across both storage backends:
- KV-backed Durable Objects
- SQLite-backed Durable Objects

### Impact

This reduces API overhead for cleanup operations and simplifies code for developers managing Durable Object lifecycles. It's particularly useful for batch cleanup, object expiration, and garbage collection patterns.

---

### Stream - Stream live inputs can now be disabled and enabled
- Published: 2026-02-24T00:00:00+00:00
- Products: Cloudflare
- Tags: feature, api
- URL: https://standarddb.com/cloudflare/cloudflare-stream-adds-ability-to-disable-and-enable-live-inputs--bpxcmttv
- Original: https://developers.cloudflare.com/changelog/post/2026-02-24-disable-live-inputs

## New Live Input Control

Cloudflare Stream now supports disabling and enabling live inputs, giving you granular control over your broadcast infrastructure. When a live input is disabled, any incoming RTMPS and SRT connection attempts will be rejected and fail to connect.

## Use Cases

This feature enables several important workflows:

- **Temporarily pause inputs** without deleting them entirely, preserving configuration and history
- **Programmatically end creator broadcasts** by disabling inputs on demand
- **Prevent unauthorized broadcasts** by disabling specific inputs when not needed
- **Manage broadcast schedules** by enabling/disabling inputs as needed

## How to Use

**Via API:** Update a live input by setting the `enabled` property to `false`:

```bash
curl --request PUT \
https://api.cloudflare.com/client/v4/accounts/{account_id}/stream/live_inputs/{input_id} \
--header "Authorization: Bearer <API_TOKEN>" \
--data '{"enabled": false}'
```

**Via Dashboard:** Toggle inputs directly from the **Live inputs** list page or the live input detail page.

All existing live inputs remain enabled by default. For detailed implementation guidance, refer to [Start a live stream](https://developers.cloudflare.com/stream/stream-live/start-stream-live/) in the Cloudflare documentation.

---

### Pipelines, Workers - Dropped event metrics, typed Pipelines bindings, and improved setup
- Published: 2026-02-24T00:00:00+00:00
- Products: Cloudflare Workers, Cloudflare
- Tags: feature, api, platform
- URL: https://standarddb.com/cloudflare/cloudflare-pipelines-adds-dropped-event-metrics-typed-bindings-and-streamlined--zfkwysqs
- Original: https://developers.cloudflare.com/changelog/2026-02-24-typed-bindings-setup-improvements-error-metrics

## Dropped Event Metrics

Cloudflare Pipelines now provides visibility into why streaming events get dropped during ingestion. When events fail to match the expected schema during delivery to sinks, the new dashboard displays errors grouped by type with specific details including `missing_field`, `type_mismatch`, `parse_failure`, and `null_value` errors.

In addition to the dashboard, dropped events can be queried programmatically via the new `pipelinesUserErrorsAdaptiveGroups` GraphQL dataset, enabling developers to:

- Query failures by error type and time range
- Aggregate error counts across specific dimensions
- Trace data quality issues back to their source
- Build custom monitoring and alerting systems

## Typed Pipeline Bindings

Previously, Worker-to-Pipeline bindings used a generic `Pipeline<PipelineRecord>` type that only surfaced schema mismatches as runtime errors. The `wrangler types` command now generates **schema-specific TypeScript types** for your Pipeline bindings, catching missing required fields and incorrect field types at compile time before deployment.

This approach mirrors the benefits of static type checking for database schemas, allowing developers to:

- Prevent schema violations before code deployment
- Leverage IDE autocomplete for field names and types
- Reduce data quality issues caused by type mismatches

## Improved Setup Experience

The `wrangler pipelines setup` command now includes a **Simple setup mode** that reduces configuration friction. Instead of manually creating R2 buckets, enabling R2 Data Catalog, generating API tokens, and configuring format/compression policies, the simplified mode:

- Applies recommended defaults automatically
- Creates R2 buckets on demand if they don't exist
- Enables R2 Data Catalog with a single command
- Provides inline validation and retry prompts instead of requiring a full restart

This significantly accelerates the onboarding experience for new Pipelines users.

---

### GitHub Enterprise Server 3.20 release candidate is available
- Published: 2026-02-24T00:00:00+00:00
- Products: GitHub
- Tags: release, feature, security, api, platform
- URL: https://standarddb.com/github/github-enterprise-server-3-20-rc-adds-immutable-releases-enhanced-secret--whtqxqwv
- Original: https://github.blog/changelog/2026-02-24-github-enterprise-server-3-20-release-candidate-is-available

## Key Features in GHES 3.20

GitHub Enterprise Server 3.20 RC is now available for testing, bringing significant improvements to deployment efficiency, monitoring, security, and policy management. Organizations can now test these features early and provide feedback before the official release.

### Release Immutability and Supply Chain Protection

**Immutable releases** are now generally available on GitHub Enterprise Server, allowing administrators to lock release assets after publication. This prevents assets from being added, modified, or deleted, and protects release tags from being moved or deleted—providing critical protection against supply chain attacks. Release attestations remain exclusive to GitHub.com.

### Enhanced Secret Scanning Capabilities

Secret scanning receives substantial improvements across multiple fronts:

- **Validity checks** now indicate whether secrets are still active, helping teams prioritize remediation efforts
- **Enterprise-level push protection bypass controls** enable centralized management of bypass permissions
- **Alert assignment** allows teams to collaborate more effectively on addressing credential leaks
- **Expanded default coverage** blocks additional secret types by default, reducing credential leak risks during pushes
- **New and improved detectors** for various secret types strengthen overall detection capabilities

Enterprise admins can make validity checks available to repository admins through the Management Console.

### Enterprise Team Management and Governance

Enterprise owners can now create and manage **enterprise teams** directly through the API or enterprise settings UI. This enables:

- Assignment of enterprise teams to organizations
- Creation and assignment of custom enterprise roles
- Management of team and user role assignments at the enterprise level
- Addition of enterprise teams to ruleset bypass lists

The feature is in public preview with some product limitations.

### New Security Manager Role

For GitHub Advanced Security customers, the new **Enterprise Security Manager role** simplifies security policy and alert management across enterprises (up to 15,000 organizations). This role is currently in public preview.

### Backup Service Availability

The **backup service**, previously in public preview, is now generally available. This managed, built-in service provides an alternative to separate backup utilities and eliminates the need for a dedicated backup host. Note that `backup-utils` will be retired starting in GHES 3.22.

### Additional UX Improvements

The improved pull request merge experience—featuring grouped status checks and better accessibility—is now generally available on GHES 3.20.

## Getting Started

Download the [3.20 release candidate](https://enterprise.github.com/releases/3.20.0/download) to test these features. Review the full [release notes](https://docs.github.com/enterprise-server@3.20/admin/release-notes) for detailed information. Organizations are encouraged to [contact GitHub support](https://support.github.com/features/enterprise-administrators-server) with feedback or questions.

---

### Pipelines, Workers - Dropped event metrics, typed Pipelines bindings, and improved setup
- Published: 2026-02-24T00:00:00+00:00
- Products: Cloudflare Workers, Cloudflare
- Tags: feature, api, platform, release
- URL: https://standarddb.com/cloudflare/cloudflare-pipelines-adds-dropped-event-metrics-typed-bindings-and-simplified--vwggtbbt
- Original: https://developers.cloudflare.com/changelog/post/2026-02-24-typed-bindings-setup-improvements-error-metrics

## Three Major Improvements for Pipelines

Cloudflare is shipping three significant updates to its Pipelines service, which ingests streaming data via Workers or HTTP endpoints, transforms it with SQL, and writes to R2 as Apache Iceberg tables.

### Dropped Event Metrics & Observability

Stream events that don't match the expected schema are now surfaced with detailed error metrics. When ingestion occurs but events fail during delivery to the sink, a new dashboard tab displays dropped events grouped by error type: `missing_field`, `type_mismatch`, `parse_failure`, or `null_value`.

Beyond the UI, you can query dropped events programmatically via the new `pipelinesUserErrorsAdaptiveGroups` GraphQL dataset, allowing you to integrate error tracking into your monitoring systems and trace issues back to the source.

### Typed Pipeline Bindings for Compile-Time Safety

Previously, sending data from Workers to Pipelines used a generic `Pipeline<PipelineRecord>` type, meaning schema mismatches were only caught at runtime as dropped events. Running `wrangler types` now generates **schema-specific TypeScript types** for your Pipeline bindings, enabling TypeScript to catch missing required fields and incorrect types at compile time before deployment.

### Simplified Setup Workflow

The new `wrangler pipelines setup` command introduces a **Simple setup mode** that applies recommended defaults and automatically creates the R2 bucket and enables R2 Data Catalog if needed. Previously, this required multiple manual steps. Validation errors during setup now prompt inline retries rather than restarting the entire process.

For detailed walkthroughs and GraphQL query examples, refer to the [Pipelines documentation](https://developers.cloudflare.com/pipelines/).

---

### Cloud Agents with Computer Use
- Published: 2026-02-24T00:00:00+00:00
- Products: 
- Tags: feature, release, platform
- URL: https://standarddb.com/cursor/cursor-adds-computer-use-to-cloud-agents-enabling-self-testing-and-demo--mjpzymdl
- Original: https://cursor.com/changelog/02-24-26

## Computer Use for Cloud Agents

Cursor has expanded its cloud agents with a new computer use capability that allows agents to directly interact with the software they build. When agents onboard onto your codebase, they now run in isolated virtual machines with full development environments, enabling them to test changes in real-time without manual intervention.

## Key Capabilities

- **Automated Testing & Demos**: Agents can execute and test the code they create, generating video recordings and screenshots to demonstrate their work
- **Rich Artifacts**: Cloud agents produce merge-ready pull requests bundled with comprehensive artifacts (videos, screenshots, and logs) for rapid review and validation
- **Multi-Platform Access**: The feature is integrated across Cursor's entire ecosystem including web, desktop, mobile, Slack, and GitHub integrations

## Getting Started

To experience cloud agents with computer use, visit [cursor.com/onboard](http://cursor.com/onboard) to watch the agent configure itself and record an automated demo. Additional details are available in Cursor's [announcement blog post](https://cursor.com/blog/agent-computer-use).

---

### 2.1.52
- Published: 2026-02-24T00:00:00+00:00
- Products: Claude Code
- Tags: release, bugfix, feature, performance, security
- URL: https://standarddb.com/anthropic/claude-code-2-1-52-fixes-extension-crash-on-windows-ships-with-50-improvements--bbfkrngj
- Original: https://github.com/anthropics/claude-code/blob/main/CHANGELOG.md

## Bug Fix

Claude Code 2.1.52 resolves a crash affecting VS Code users on Windows where the `claude-vscode.editor.openLast` command was not found, preventing the extension from functioning properly.

## Recent Improvements (2.1.50-2.1.52)

The latest releases include substantial stability and performance enhancements:

**Agent & Isolation Improvements**
- Added `isolation: "worktree"` support for agents to run in temporary git worktrees, improving sandbox isolation
- New `--worktree` flag to launch Claude in an isolated environment
- Agent background task execution with `background: true` configuration
- New `claude agents` CLI command to list configured agents

**Memory & Performance Optimizations**
- Fixed multiple memory leaks affecting long sessions (LSP diagnostics, task output, circular buffers, ChildProcess references)
- Tool results larger than 50K characters now persist to disk, reducing context window usage
- Improved startup performance for headless mode by deferring non-essential imports
- Cleared internal caches and file history snapshots to prevent unbounded memory growth

**Developer Experience**
- Enhanced MCP OAuth with step-up auth support and discovery caching
- Added `startupTimeout` configuration for LSP servers
- Support for custom npm registries and version pinning when installing plugins
- Improved error messages suggesting corrected file paths
- `/model` picker now shows human-readable labels (e.g., "Sonnet 4.5") with upgrade hints

**Security & Reliability**
- Fixed security issue where `statusLine` and `fileSuggestion` hook commands could execute without workspace trust
- Fixed WebSocket reconnect bug causing duplicate `control_response` messages and API 400 errors
- Fixed session visibility issues with symlinked working directories
- Added environment variables (`CLAUDE_CODE_ACCOUNT_UUID`, `CLAUDE_CODE_USER_EMAIL`, `CLAUDE_CODE_ORGANIZATION_UUID`) for SDK callers

**Plugin & Tool System**
- `WorktreeCreate` and `WorktreeRemove` hook events for custom VCS setup/teardown
- Plugins can now ship `settings.json` for default configuration
- Fixed slash command autocomplete crash with YAML array descriptions in plugin SKILL.md
- BashTool skips login shell by default when snapshots are available, improving command execution performance

**Linux Compatibility**
- Fixed native modules not loading on systems with glibc older than 2.30 (e.g., RHEL 8)

---

### Using NVFP4 Low-Precision Model Training for Higher Throughput Without Losing Accuracy
- Published: 2026-02-23T18:00:00+00:00
- Products: 
- Tags: feature, performance, model, sdk, open-source
- URL: https://standarddb.com/nvidia/nvidia-achieves-1-6x-training-throughput-gains-with-nvfp4-low-precision-format--pncdphry
- Original: https://developer.nvidia.com/blog/using-nvfp4-low-precision-model-training-for-higher-throughput-without-losing-accuracy

## NVFP4 Low-Precision Training Now Production-Ready

NVIDIA has published comprehensive benchmarks showing that NVFP4, a 4-bit floating-point format, achieves production-ready performance for large-scale model training. The research compares three low-precision approaches—FP8-CS, MXFP8, and NVFP4—against standard BF16 precision training to address critical challenges in scaling transformer models: training throughput, memory constraints, and computational costs.

## Key Results and Performance Gains

Experiments on Llama 3 8B and NVIDIA's Research-8B models (trained on 1 trillion tokens) demonstrate:

- **Up to 1.6x throughput improvement** with NVFP4 compared to BF16
- **Pretraining and downstream benchmark accuracy nearly identical** to BF16, with NVFP4 requiring selective BF16 layers for convergence stability
- **Significant memory savings** enabling larger micro-batch sizes and improved scalability
- MXFP8 slightly outperforms standard FP8-CS due to block-level scaling optimized for NVIDIA Blackwell architecture

## How NVFP4 Works

NVFP4 reduces memory bandwidth and computational demand by using 4-bit numerical formats for weights and activations during training. Unlike simpler approaches, it employs a hierarchical two-level scaling strategy to balance numerical accuracy with performance, allowing GPUs to process more operations per cycle and substantially increase training throughput.

## Available Today via NeMo Megatron Bridge

The production-ready recipes are available through **NVIDIA NeMo Megatron Bridge**, an open-source library that is part of the broader NVIDIA NeMo framework. Developers can switch between precision formats with minimal code modifications, making it straightforward to adopt these efficiency gains in real-world training pipelines.

## What Developers Should Know

If you're training large language models, NVFP4 and other low-precision formats on B200 GPUs offer a direct path to faster training with lower memory requirements and reduced costs, without sacrificing model quality on standard benchmarks. The approach is production-ready and can be integrated into existing training workflows using NeMo Megatron Bridge.

---

### Support for now.json will be removed on March 31, 2026
- Published: 2026-02-23T13:00:00+00:00
- Products: Vercel
- Tags: deprecation, platform
- URL: https://standarddb.com/vercel/vercel-deprecating-now-json-config-files-migration-deadline-set-for-march-31--wrrvfbgq
- Original: https://vercel.com/changelog/support-for-now-json-will-be-removed-on-march-31-2026

## End of Life for now.json

Vercel has announced the deprecation of the legacy `now.json` configuration file format. Support will be officially removed on **March 31, 2026**, giving developers over a year to complete their migration.

### Migration Path

Migrating from `now.json` to the newer configuration format is straightforward:
- **Rename** your `now.json` file to `vercel.json`
- **No content changes** are required—the file content remains the same
- All existing configurations will continue to work with the new filename

### What Developers Should Know

This deprecation is part of Vercel's effort to modernize its configuration system. The `vercel.json` format is now the standard for file-based project configuration.

For teams with more complex configuration requirements, Vercel also offers **`vercel.ts`**, which enables programmatic project configuration for advanced use cases.

### Next Steps

Developers should plan to update any existing projects still using `now.json` before the March 31, 2026 deadline. Refer to the [official documentation](https://vercel.com/docs/project-configuration#file-based-configuration) for details on configuring projects with `vercel.json`.

---

### Introducing npm i chat – One codebase, every chat platform
- Published: 2026-02-23T13:00:00+00:00
- Products: Vercel AI SDK, Next.js, Vercel
- Tags: release, feature, sdk, open-source, platform, integration
- URL: https://standarddb.com/vercel/vercel-opens-chat-sdk-in-public-beta-unifies-bot-development-across-slack-teams--dkmrgnbq
- Original: https://vercel.com/changelog/chat-sdk

## Unified Chatbot Development Across Platforms

Building chatbots that work across multiple communication platforms has historically required maintaining separate codebases for each service's API. Vercel's new Chat SDK addresses this fragmentation by providing a single TypeScript library that abstracts away platform-specific differences while maintaining native functionality on each platform.

## Key Features and Architecture

The Chat SDK is built on an event-driven architecture with type-safe handlers for common chatbot interactions:

- **Event Handlers**: Built-in support for mentions, messages, reactions, button clicks, and slash commands
- **Native UI Components**: JSX-based card and modal components that render natively on each platform rather than emulating a single design
- **State Management**: Pluggable adapters for Redis, ioredis, and in-memory storage to handle distributed state
- **Stream Support**: Integrated with the Vercel AI SDK to enable real-time streaming of AI responses directly to chat platforms
- **Platform Coverage**: Initial support for Slack, Microsoft Teams, Google Chat, Discord, GitHub, and Linear

## Getting Started and Deployment

The SDK follows a modular architecture with a core `chat` package and separate platform adapters. Developers can start with simple mention-based interactions and scale to complex multi-step workflows. Vercel provides guides for common patterns including Slack bots with Next.js, Discord support bots with Nuxt, GitHub bots with Hono, and automated code review workflows.

The Chat SDK is available in public beta today with full documentation at chat-sdk.dev. Developers can start building immediately with a free Vercel account.


---

### v0.5.99
- Published: 2026-02-23T08:33:30+00:00
- Products: Sim Studio
- Tags: release, feature, bugfix
- URL: https://standarddb.com/sim/sim-v0-5-99-adds-expandable-workflow-blocks-and-fixes-parallel-subflow-display--cvymdwnj
- Original: https://github.com/simstudioai/sim/releases/tag/v0.5.99

## What's New

**Expandable Child Workflow Blocks**: The console terminal now supports expandable child workflow blocks, allowing developers to inspect nested workflows directly in the terminal interface. This enhancement improves visibility and debugging capabilities when working with complex workflow hierarchies.

## Bug Fixes

- **Parallel Subflow Display**: Corrected active state pulsing and duration display for parallel subflow blocks, ensuring accurate visual feedback during parallel execution
- **Security Update**: Resolved a security restriction to allow HTTP connections for localhost and loopback addresses, enabling local development workflows without requiring HTTPS

## Technical Details

This is a patch release (v0.5.99) focused on developer experience improvements and bug fixes. The changes make it easier to work with nested workflows in the console and ensure proper visualization of parallel execution states.

---

### Cloudflare One is the first SASE offering modern post-quantum encryption across the full platform
- Published: 2026-02-23T06:00:00+00:00
- Products: Cloudflare
- Tags: release, security, feature, platform
- URL: https://standarddb.com/cloudflare/cloudflare-one-becomes-first-sase-platform-with-post-quantum-encryption-across--cmbjcyff
- Original: https://blog.cloudflare.com/post-quantum-sase

## Post-Quantum Encryption Now Available Across Cloudflare One

Cloudflare has extended post-quantum cryptographic protection to its complete Secure Access Service Edge (SASE) platform. Building on its earlier launch of post-quantum Secure Web Gateway and Zero Trust solutions during Security Week 2025, the company now offers standards-compliant post-quantum hybrid ML-KEM (Module-Lattice-based Key-Encapsulation Mechanism) across all major on-ramps and off-ramps.

### What's New

**Cloudflare IPsec WAN Protection** — Post-quantum encryption is now available for Cloudflare IPsec, the cloud-native WAN-as-a-Service that establishes encrypted tunnels from customer networks to Cloudflare's global infrastructure. This service currently operates in closed beta; organizations can request access via Cloudflare's beta signup page.

**Cloudflare One Appliance Support** — The physical or virtual WAN appliance now includes post-quantum encryption support as of version 2026.2.0, which is generally available today. The appliance establishes Cloudflare IPsec connections and simplifies configuration while providing automatic failover to healthy data centers.

### Why This Matters

**Regulatory Deadline Approaching** — The National Institute of Standards and Technology (NIST) has set a 2030 deadline for deprecating RSA and Elliptic Curve Cryptography (ECC) in favor of post-quantum alternatives. Organizations that delay migration risk non-compliance as the deadline nears.

**Crypto Agility** — Modern enterprises need the ability to swap cryptographic algorithms without extensive system overhauls. Historical precedent shows that deprecating cryptography takes decades (MD5 caused problems 20 years after deprecation). Built-in post-quantum support within Cloudflare One simplifies remote access and site-to-site connectivity provisioning.

**Harvest Now, Decrypt Later** — Adversaries are already collecting encrypted network traffic to decrypt once quantum computers become powerful enough. Organizations with data sensitive beyond a few years (financial records, health information, state secrets) face present risk without post-quantum protection.

### Technical Details

The ML-KEM standard has achieved broad industry adoption for TLS key agreement, typically deployed alongside classical ECDHE to create "hybrid ML-KEM"—a dual-algorithm approach with no reduction in security. Over 60% of human-generated TLS traffic to Cloudflare's network now uses hybrid ML-KEM, demonstrating adoption viability. The hybrid approach offers minimal performance impact and requires no specialized hardware or physical infrastructure.

---

### Enterprise team support in organization APIs
- Published: 2026-02-23T00:00:00+00:00
- Products: GitHub
- Tags: api, feature, platform
- URL: https://standarddb.com/github/github-adds-enterprise-team-support-to-organization-apis-unifying-team-discovery--qzmnwmdg
- Original: https://github.blog/changelog/2026-02-23-enterprise-team-support-in-organization-apis

## Enterprise Teams Now Unified in Organization APIs

GitHub has extended several organization API endpoints to support enterprise teams alongside existing organization teams. This change eliminates the previous complexity where administrators needed to query separate endpoints and manage different permission levels to get a complete view of all teams in their organization.

The following endpoints now support enterprise teams:

- **Teams** - Query and manage teams with enterprise team support
- **Team members** - Interact with team membership across both team types
- **Organization repository rulesets** - Apply rules using enterprise or organization teams
- **Repository rulesets** - Configure repository-level rules with unified team support

## Key Capabilities

With these API updates, organization owners can now:

- **Query both team types in a single request** - Retrieve enterprise and organization teams without multiple API calls
- **Filter by team type** - The "List Teams" endpoint supports filtering to distinguish between team scopes
- **Consistent response format** - Enterprise and organization teams are returned in a standardized format for easier integration
- **Proper permission boundaries** - Organization owners see teams scoped to their domain, while enterprise owners get full enterprise-wide visibility

## Implementation Details

When using these endpoints, invalid enterprise team inputs will return appropriate error responses. The feature is currently in public preview and subject to change based on developer feedback.

GitHub is actively soliciting feedback through the [community discussions](https://github.com/orgs/community/discussions/177040) to help shape the final implementation of this feature.

---

### Hyperdrive - Hyperdrive no longer caches queries using STABLE PostgreSQL functions
- Published: 2026-02-23T00:00:00+00:00
- Products: Cloudflare Workers, Cloudflare
- Tags: bugfix, api
- URL: https://standarddb.com/cloudflare/hyperdrive-stops-caching-queries-with-postgresql-stable-functions-to-prevent--xmpsyhbc
- Original: https://developers.cloudflare.com/changelog/post/2026-02-23-hyperdrive-stable-functions-uncacheable

## Query Caching Behavior Updated

Hyperdrive now correctly treats queries containing PostgreSQL `STABLE` functions as uncacheable, in addition to previously detected `VOLATILE` functions. This change prevents queries from returning stale or incorrect data when STABLE functions are used.

## What Changed

Previously, Hyperdrive only marked queries containing `VOLATILE` functions (such as `RANDOM()` or `LASTVAL()`) as uncacheable. `STABLE` functions like `NOW()`, `CURRENT_TIMESTAMP`, and `CURRENT_DATE` were incorrectly allowed to be cached.

Since `STABLE` functions can return different results across different SQL statements within the same transaction, caching their results could serve stale or incorrect data to your application.

## Migration Path

If your queries use `STABLE` functions and were relying on caching behavior, you'll need to adjust your approach:

- Move the function call to your application code and pass the result as a query parameter
- **Example**: Instead of `WHERE created_at > NOW()`, compute the timestamp in your Worker and pass it as `WHERE created_at > $1`

## Implementation Details

Hyperdrive uses text-based pattern matching to detect uncacheable functions. Note that references to function names (like `NOW()`) in SQL comments will also cause the entire query to be marked as uncacheable.

For more information, see the [Query caching documentation](https://developers.cloudflare.com/hyperdrive/concepts/query-caching/) and [troubleshooting guide](https://developers.cloudflare.com/hyperdrive/observability/troubleshooting/).

---

### Agents, R2, Containers - Backup and restore API for Sandbox SDK
- Published: 2026-02-23T00:00:00+00:00
- Products: Cloudflare Workers, Cloudflare
- Tags: feature, api, platform
- URL: https://standarddb.com/cloudflare/cloudflare-sandbox-sdk-adds-backup-and-restore-api-for-persistent-environments--mxjhxbrj
- Original: https://developers.cloudflare.com/changelog/2026-02-23-sandbox-backup-restore-api

## New Backup and Restore API

Cloudflare Sandbox introduces `createBackup()` and `restoreBackup()` methods for persisting and restoring sandbox environments. This feature addresses a common pain point: initial setup tasks like cloning repositories and installing dependencies can take minutes, making it impractical to repeat these steps every time a user resumes work.

## Key Capabilities

The backup and restore API provides:

- **Persist and reuse across sessions** — Store backup handles in KV, D1, or Durable Object storage for retrieval in future sessions
- **Multi-instance forking** — Deploy a single backup across many sandboxes to enable parallel work
- **Named backups** — Assign human-readable labels for easier management
- **Automatic cleanup** — Configure TTLs to automatically remove backups after they're no longer needed

## Storage and Configuration

Backups are stored in [R2](https://developers.cloudflare.com/r2), Cloudflare's object storage service. Developers can leverage [R2 object lifecycle rules](https://developers.cloudflare.com/sandbox/guides/backup-restore/#configure-r2-lifecycle-rules-for-automatic-cleanup) to control backup retention and prevent indefinite storage costs.

## Getting Started

The [backup and restore guide](https://developers.cloudflare.com/sandbox/guides/backup-restore/) provides setup instructions and usage patterns, while the [Backups API reference](https://developers.cloudflare.com/sandbox/api/backups/) documents all available methods. A typical workflow involves creating a backup after initial setup, storing the handle for later use, then restoring instead of re-running expensive initialization steps.



---

### Agents, R2, Containers - Backup and restore API for Sandbox SDK
- Published: 2026-02-23T00:00:00+00:00
- Products: Cloudflare Workers
- Tags: feature, api, platform
- URL: https://standarddb.com/cloudflare/cloudflare-sandbox-sdk-adds-backup-and-restore-api-for-point-in-time-snapshots--yjbhgjvz
- Original: https://developers.cloudflare.com/changelog/post/2026-02-23-sandbox-backup-restore-api

## Sandbox Backup and Restore API Now Available

Cloudflare has introduced backup and restore capabilities to the Sandbox SDK, enabling developers to create point-in-time snapshots of directories and restore them across sessions. This addresses a critical pain point: expensive setup operations like `git clone` and `npm install` that can take minutes and must be repeated every time a user starts a new sandbox environment.

### Key Capabilities

The new API provides:

- **Session persistence** — Store backup handles in KV, D1, or Durable Objects to reuse environments across sessions
- **Multi-instance support** — Fork a single backup across many sandboxes for parallel work
- **Named backups** — Add human-readable labels for easier management and tracking
- **Automatic cleanup** — Configure time-to-live (TTL) durations so backups are automatically removed from R2 storage

### How It Works

After completing initial setup steps (cloning a repository, installing dependencies, etc.), developers call `createBackup()` to create a snapshot. The backup handle can be stored and later passed to `restoreBackup()` to resume work without repeating setup. Backups are stored in R2 and support [R2 object lifecycle rules](https://developers.cloudflare.com/sandbox/guides/backup-restore/#configure-r2-lifecycle-rules-for-automatic-cleanup) for automatic cleanup.

### Getting Started

Developers should consult the [backup and restore guide](https://developers.cloudflare.com/sandbox/guides/backup-restore/) for setup instructions and usage patterns, or the [Backups API reference](https://developers.cloudflare.com/sandbox/api/backups/) for complete method documentation.

---

### IP allow list coverage extended to EMU namespaces in public preview
- Published: 2026-02-23T00:00:00+00:00
- Products: GitHub
- Tags: feature, security, platform
- URL: https://standarddb.com/github/github-extends-ip-allow-list-coverage-to-enterprise-managed-users-namespaces-in--mlpxtrpj
- Original: https://github.blog/changelog/2026-02-23-ip-allow-list-coverage-extended-to-emu-namespaces-in-public-preview

## IP Allow List Expansion for EMU Namespaces

GitHub Enterprise Cloud has extended its native IP allow list configuration to cover user namespaces associated with Enterprise Managed Users (EMUs). This new public preview feature ensures that enterprises can enforce consistent network access policies across all repositories within their organization, including those owned by user accounts managed through their EMU system.

### What's New

The IP allow list now filters access to enterprise-managed user namespaces across multiple access vectors:

- **Web UI access**: All browser-based interactions are subject to IP allow list restrictions
- **Git protocol**: Command-line git operations (push, pull, fetch) are filtered
- **API access**: All API requests are evaluated against the IP allow list
- **All credential types**: Personal access tokens, GitHub App tokens, and SSH keys are all covered by the policy

### Key Benefits

This enhancement strengthens security posture for enterprises that need consistent network access controls. By extending IP allow list policies to EMU user namespaces, organizations can:

- Enforce enterprise-wide network security policies uniformly
- Ensure repositories owned by managed users adhere to the same access restrictions
- Maintain consistent security standards across all enterprise resources

### Getting Started

Enterprises interested in enabling this feature can review the [GitHub documentation on configuring IP allow lists](https://docs.github.com/enterprise-cloud@latest/admin/configuring-settings/hardening-security-for-your-enterprise/restricting-network-traffic-to-your-enterprise-with-an-ip-allow-list) for detailed setup and configuration guidance.

---

### Safely inject credentials in HTTP headers with Vercel Sandbox
- Published: 2026-02-23T00:00:00+00:00
- Products: Vercel
- Tags: feature, security, api
- URL: https://standarddb.com/vercel/vercel-sandbox-adds-secure-header-injection-for-api-credentials-in-untrusted--kwrmyztw
- Original: https://vercel.com/changelog/safely-inject-credentials-in-http-headers-with-vercel-sandbox

## Secure Credential Handling for Sandboxed Code

Vercel Sandbox now includes a credential injection feature that automatically adds HTTP headers to outbound requests from sandboxed code. This keeps API keys and authentication tokens outside the sandbox virtual machine, preventing malicious or compromised code from exfiltrating credentials.

## How It Works

Header injection is configured via the `transform` property in the sandbox's network policy. When the sandbox makes an HTTPS request to a matching domain, Vercel's firewall automatically injects or replaces the specified headers before forwarding the request. This is particularly valuable for AI agent workflows where prompt injection attacks could otherwise compromise credentials.

```typescript
const sandbox = await Sandbox.create({
  networkPolicy: {
    allow: {
      "ai-gateway.vercel.sh": [{
        transform: [{
          headers: {
            authorization: `Bearer ${process.env.AI_GATEWAY_API_KEY}`
          }
        }],
      }],
    },
  },
});
```

## Key Capabilities

- **Domain Matching**: Supports exact domains and wildcards (e.g., `*.github.com`) to control which requests receive injected headers
- **Header Overwriting**: Injected headers fully replace any headers set by sandbox code, preventing the sandbox from substituting its own credentials
- **Dynamic Updates**: Network policies—including injection rules—can be updated on running sandboxes without restart, enabling multi-phase workflows
- **Universal Compatibility**: Works with all egress network policy configurations, including open internet access alongside restricted domain allow-lists

## Multi-Phase Workflows

The live update capability enables secure multi-phase workflows. You can inject credentials during setup phases (e.g., cloning repositories), then lock down the network policy to `deny-all` before executing untrusted code, eliminating any risk of credential access.

## Availability

This feature is available to all Pro and Enterprise customers. For more details, see the [Vercel Sandbox documentation](https://vercel.com/docs/vercel-sandbox/concepts/firewall#credentials-brokering).

---

### Security Center - Saved views for Threat Events
- Published: 2026-02-23T00:00:00+00:00
- Products: Cloudflare
- Tags: feature, security, platform
- URL: https://standarddb.com/cloudflare/cloudflare-security-center-adds-saved-views-for-threat-events-dashboard--mgwflvjs
- Original: https://developers.cloudflare.com/changelog/post/2026-02-23-saved-views-in-threat-events

## Saved Views for Threat Events Dashboard

Cloudflare has released saved views functionality for the Threat Events dashboard within Security Center, allowing analysts to create and store custom filtered configurations.

### What's New

- **Custom view persistence**: Create multiple saved views with different filter configurations (industry-specific attacks, geographic regions, threat types, etc.) that persist across sessions
- **One-click access**: Return to complex filtered dashboards instantly without manually re-applying filters
- **Team standardization**: Share standardized views across your security team to ensure consistent analysis and reduce pattern-detection gaps

### Why This Matters

Security analysts previously had to manually reconfigure filters every time they accessed the Threat Events dashboard. Complex scenarios—combining industry datasets with geographic origin filtering, for example—required multiple steps to recreate. This update eliminates that repetitive setup overhead, enabling faster threat intelligence workflows.

### Getting Started

Saved views are available now in the Threat Intelligence section for Cloudforce One subscribers. Navigate to **Application Security > Threat Intelligence > Threat Events** in the Cloudflare dashboard to begin creating your custom views.

---

### Security Center - Saved views for Threat Events
- Published: 2026-02-23T00:00:00+00:00
- Products: Cloudflare
- Tags: feature, security
- URL: https://standarddb.com/cloudflare/cloudflare-adds-saved-views-to-security-center-s-threat-events-dashboard--zwvbblzk
- Original: https://developers.cloudflare.com/changelog/2026-02-23-saved-views-in-threat-events

## Saved Views for Threat Events

Cloudflare has released a new feature in Security Center that allows users to create and save custom configurations of the Threat Events dashboard. This enhancement addresses a key pain point for security analysts who previously had to manually re-apply complex filters every time they accessed threat intelligence data.

### What's New

- **Custom saved views** can now be created with specific filter combinations, such as industry-specific attack patterns or geographic data flows
- **One-click access** to frequently-used threat intelligence configurations without manual reconfiguration
- **Team standardization** through shared saved views, ensuring consistent data analysis across security teams

### Why It Matters

Threat intelligence analysis requires personalized, contextual data filtering. By eliminating repetitive filter setup, analysts can focus more time on actual threat investigation rather than dashboard configuration. Teams can also reduce the risk of missing critical patterns that might occur due to inconsistent filtering approaches.

### Getting Started

Cloudforce One subscribers can begin using this feature immediately in **Application Security > Threat Intelligence > Threat Events** within the Cloudflare dashboard to create and manage their custom threat event views.

---

### Enterprise-defined custom organization roles are generally available
- Published: 2026-02-23T00:00:00+00:00
- Products: GitHub
- Tags: release, feature, api
- URL: https://standarddb.com/github/github-enterprise-administrators-can-now-create-custom-organization-roles--vbmgxrmf
- Original: https://github.blog/changelog/2026-02-23-enterprise-defined-custom-organization-roles-are-generally-available

## Enterprise-Defined Custom Organization Roles Now Available

GitHub has made enterprise-defined custom organization roles generally available, enabling enterprise administrators to create standardized role definitions that apply across all organizations in their enterprise.

### Key Capabilities

- **Enterprise-wide standardization**: Create up to 20 custom organization roles at the enterprise level, available across all organizations
- **Managed by authorized admins**: Only authorized enterprise administrators can edit these roles, maintaining security and consistency
- **Flexible assignment**: Organization administrators can assign these roles through the organization settings UI or via the [organization role assignment API](https://docs.github.com/enterprise-cloud@latest/rest/orgs/organization-roles#assign-an-organization-role-to-a-team)
- **Same permissions model**: Enterprise-defined roles use the same set of organization and repository permissions as organization-level custom roles, with no functional differences

### Why This Matters

This feature addresses a key challenge for enterprises managing role-based access control (RBAC) across multiple organizations. By centralizing role definitions, enterprises can:
- Enforce consistent compliance requirements
- Reduce administration overhead
- Make it easier for users to move between organizations with familiar roles
- Maintain security by ensuring only authorized administrators can modify these roles

### Getting Started

Learn more about [creating enterprise-defined custom organization roles](https://docs.github.com/enterprise-cloud@latest/admin/managing-accounts-and-repositories/managing-roles-in-your-enterprise/create-custom-roles#creating-organization-custom-roles) in the GitHub documentation, and share feedback in the [GitHub Community Enterprise forum](https://github.com/orgs/community/discussions/categories/enterprise).

---

### Docker Desktop 4.62.0
- Published: 2026-02-23T00:00:00+00:00
- Products: 
- Tags: release, feature, security, bugfix, platform
- URL: https://standarddb.com/docker/docker-desktop-4-62-0-adds-mcp-server-profiles-and-custom-catalogs--klxhrtlq
- Original: https://docs.docker.com/desktop/release-notes

## Key Features

**MCP Toolkit Enhancements**: Docker Desktop 4.62.0 adds support for organizing Model Context Protocol (MCP) servers through profiles — named collections that help you manage and deploy MCP servers more effectively. Additionally, you can now create custom catalogs — curated collections of MCP servers tailored for your team or organization.

## Updates and Improvements

- **Linux Kernel**: Updated to v6.12.69 for improved stability and performance
- **Docker Model Runner**: Added support for vLLM Metal on Mac with Apple chips, enabling optimized model inference on Apple silicon
- **Bug Fixes**: Fixed an issue where background update checks weren't respecting the "Automatically check for updates" setting when disabled

## Platform-Specific Fixes

**Linux**: Resolved a networking crash that occurred with QEMU 10.2.0 and later versions.

## Security

This release addresses **CVE-2026-2664**, an out-of-bounds read vulnerability in the grpcfuse kernel module. Users should update to this version to receive the security patch.

## Availability

Docker Desktop 4.62.0 is available for Windows (Intel and ARM), Mac (Apple and Intel chips), and Linux (Debian, RPM, and Arch distributions). Gradual rollout is in progress — if the update isn't available yet, check back within a week.

---

### Databricks SQL version 2025.40 is rolling out in Current
- Published: 2026-02-23T00:00:00+00:00
- Products: 
- Tags: release, feature, api
- URL: https://standarddb.com/databricks/databricks-sql-2025-40-introduces-sql-scripting-ga-and-expanded-parameter--cvjjfjsw
- Original: https://docs.databricks.com/aws/en/sql/release-notes/2026

## SQL Scripting Now Generally Available

SQL scripting is now generally available in Databricks SQL, enabling developers to write procedural logic directly in SQL. This includes conditional statements, loops, local variables, and exception handling—expanding SQL beyond traditional declarative queries into a full procedural language.

## Enhanced Parameterization Support

**Parameter markers** are now supported in nearly all SQL contexts where literal values are allowed:
- DDL statements: `CREATE VIEW v AS SELECT ? AS c1`
- Column type definitions: `DECIMAL(:p, :s)`
- Comments: `COMMENT ON t IS :comment`

The **IDENTIFIER clause** has been expanded to support dynamic object naming in additional contexts, including column aliases and definitions: `AS IDENTIFIER(:name)` and `IDENTIFIER(:pk) BIGINT NOT NULL`.

**Literal string coalescing** now works everywhere string literals are permitted, allowing sequential strings like `'Hello' ' World'` to automatically combine into `'Hello World'`.

## New Approximate Computation Functions

Two major function libraries have been added for statistical operations:

- **Theta Sketch functions** for approximate distinct counts and set operations: `theta_sketch_agg`, `theta_union_agg`, `theta_intersection_agg`, and supporting functions
- **KLL Sketch functions** for approximate quantile computation with variants for `BIGINT`, `DOUBLE`, and `FLOAT` types, including aggregation, merging, and retrieval functions

Additionally, a new **BITMAP_AND_AGG** function complements the existing bitmap function library for bit-level operations.

## Action Items

Existing SQL code can take advantage of parameterization features to improve security and reusability. Teams working with approximate analytics should evaluate the new Theta Sketch and KLL Sketch functions for performance optimization on large datasets.

---

### v0.5.98
- Published: 2026-02-22T02:15:41+00:00
- Products: Sim Studio
- Tags: release, feature, bugfix, platform, integration
- URL: https://standarddb.com/sim/sim-studio-v0-5-98-adds-hex-tool-support-and-oauth-client-metadata-discovery--vppvzmtv
- Original: https://github.com/simstudioai/sim/releases/tag/v0.5.98

## Release Overview

Sim Studio v0.5.98 introduces new capabilities and important stability improvements. The release includes two notable features alongside several bug fixes that address infrastructure and model management concerns.

## New Features

**Hex Tool Integration**: A new hex tool has been added to the platform, expanding the available capabilities for workflows and agents.

**OAuth Enhancement**: Support for CIMD (Client Information Meta Data) has been integrated, improving the OAuth client metadata discovery process for better integration flexibility.

## Bug Fixes & Stability

- **Redis Resilience**: Fixed an issue preventing false rate limits and code execution failures during Redis outages, ensuring more reliable operation when infrastructure components experience temporary issues
- **Model Updates**: Removed the retired Claude 3.7 Sonnet model and updated default model references accordingly
- **Scope Parameter Fixes**: Corrected scope parameter renames within hex operations
- **Deployment Improvements**: Enhanced the change detection hook by reusing the subblock merge helper for more efficient deployments

## What Developers Should Know

The Claude 3.7 Sonnet removal may require updates to workflows using this model—check your configurations if you're relying on this deprecated model. The improved Redis outage handling makes the platform more resilient to infrastructure hiccups. The new hex tool and OAuth CIMD support expand integration possibilities for workflow builders.

---

### v0.5.97
- Published: 2026-02-21T10:14:01+00:00
- Products: Sim Studio
- Tags: release, bugfix, integration
- URL: https://standarddb.com/sim/sim-studio-v0-5-97-fixes-oauth-metadata-url-discovery-in-mcp--crqgnkbg
- Original: https://github.com/simstudioai/sim/releases/tag/v0.5.97

## Bug Fix

**OAuth Discovery Metadata URL Handling**: This release resolves an issue in the MCP (Model Context Protocol) implementation where OAuth discovery metadata URLs were not properly utilizing the base URL configuration. The fix ensures that `getBaseUrl` is correctly applied to OAuth discovery metadata endpoints.

## What Developers Need to Know

If you're using Sim Studio with MCP integrations that rely on OAuth authentication, this patch improves the reliability of OAuth discovery flows. The fix applies to scenarios where OAuth metadata endpoints need to respect custom base URL configurations, which is particularly important in enterprise or restricted network environments.

## Getting the Update

Sim Studio v0.5.97 is available now. Users should update to benefit from improved OAuth integration stability.

---

### v0.17.0-rc1: mlx: remove noisy error output from dynamic library loading (#14346)
- Published: 2026-02-21T07:46:07+00:00
- Products: 
- Tags: release, bugfix
- URL: https://standarddb.com/ollama/ollama-v0-17-0-rc1-removes-noisy-error-output-from-mlx-library-loading--jwqlnsvc
- Original: https://github.com/ollama/ollama/releases/tag/v0.17.0-rc1

## MLX Dynamic Library Loading Improved

This release candidate addresses a user-facing issue introduced in a recent optimization that attempts to load `libmlxc.dylib` via rpath before searching directories. This rpath-first approach benefits Homebrew installations where rpath is correctly configured.

## The Problem

When rpath isn't set—the common case for app bundle installations—the dynamic library loading would fail and print a misleading error message to stderr:

```
ERROR - dynamic.c:21 - CHECK failed: handle->ctx != NULL
```

Users seeing this error message were confused because the library actually loaded successfully via the fallback path. The error was an expected part of the loading sequence, not an actual failure.

## The Solution

The fix replaces the overly aggressive CHECK macro with a simple return code, allowing the C code to fail silently when the rpath-based load attempt doesn't work. The Go code already handles logging appropriately:

- **tryLoadByName()** now fails silently (intentional fallback behavior)
- **tryLoadFromDir()** still logs errors via slog.Error() when explicit path loading actually fails

This preserves proper error diagnostics for real failures while eliminating noise from expected fallback scenarios.

---

### v0.5.96
- Published: 2026-02-21T02:34:12+00:00
- Products: Sim Studio
- Tags: release, feature, bugfix, integration, api
- URL: https://standarddb.com/sim/sim-v0-5-96-adds-slack-ephemeral-messages-and-oauth-2-1-support--szgybyxl
- Original: https://github.com/simstudioai/sim/releases/tag/v0.5.96

## New Features

Sim Studio v0.5.96 introduces enhanced Slack integration and authentication improvements:

**Slack Ephemeral Messages**: A new ephemeral message send tool is now available, allowing workflows to send temporary messages visible only to specific Slack users. This update includes refreshed CI pipelines and updated documentation.

**OAuth 2.1 Provider**: Added OAuth 2.1 provider support for MCP (Model Context Protocol) connector authentication, enabling more secure and modern authentication flows for integrations.

## Bug Fixes

This release addresses several stability issues:

- **Copilot**: Fixed handling of negated operation conditions in block configuration extraction, improving workflow logic evaluation
- **Slack Events**: Corrected payload handling for `reaction_added` and `reaction_removed` events, ensuring triggers fire correctly
- **Logs**: Replaced stale `initialData` with `placeholderData` in log details views for more accurate logging display

## Getting Started

Update to v0.5.96 to access ephemeral Slack messaging and the new OAuth 2.1 authentication option for your MCP connectors. All improvements are backward compatible with existing workflows.

---

### Granola + HubSpot integration: automatic CRM updates from meeting notes
- Published: 2026-02-21T00:00:00+00:00
- Products: 
- Tags: integration, feature, api
- URL: https://standarddb.com/granola/granola-launches-native-hubspot-integration-for-automatic-crm-syncing-from--vbfhqqzw
- Original: https://www.granola.ai/blog/granola-hubspot-integration-crm-updates

## Native HubSpot Integration

Granola has released a native integration with HubSpot that eliminates manual CRM data entry after customer meetings. The integration is available on Business and Enterprise plans, allowing users to push enhanced meeting notes directly to HubSpot Contact records in a single click. The sync includes meeting titles, dates, participant lists, enhanced note bodies, and extracted action items.

## Key Features

**Automatic note enrichment:** Granola's AI enhancement automatically adds relevant context and quotes from meeting transcripts before syncing to HubSpot. This means your CRM receives polished summaries, not just raw notes.

**Attendee-based matching:** The integration intelligently matches notes to the correct HubSpot Contact records using email addresses from your calendar invite, reducing manual selection.

**Custom template support:** If you use custom note templates in Granola, those formatted sections carry over as structured text in HubSpot activity logs.

## Setup and Prerequisites

The integration requires:
- A Granola Business or Enterprise plan ($14/user/month minimum)
- A work email account (Google Workspace or Microsoft 365, not personal Gmail/Outlook)
- HubSpot Super Admin or App Marketplace access
- A synced calendar for accurate attendee data

Setup takes three steps: connect your HubSpot account via OAuth, share a note to HubSpot from the Share menu, and verify the synced note appears in the Contact's activity timeline.

## Zapier Alternative for Automation

For fully automatic syncing without manual share steps, Granola supports Zapier integration with triggers for new notes or notes added to specific folders. Zapier enables advanced routing (notes to multiple HubSpot objects), folder-based automation, and integration with other tools in your workflow.

## Use Cases

**Sales teams** benefit from pipeline context automatically captured during discovery calls—deal signals, budget info, and next steps appear on Contact records without manual CRM entry. **Research and product teams** can now share customer insights and feature requests from call notes directly to shared Contact records, making discoveries visible to the entire team rather than trapped in personal notes folders.

This integration directly addresses the gap between meeting intelligence (captured in transcripts and notes) and CRM visibility, ensuring customer conversations become searchable institutional knowledge rather than siloed information.

---

### v0.5.95
- Published: 2026-02-20T22:20:30+00:00
- Products: Sim Studio
- Tags: release, feature, integration, model, bugfix
- URL: https://standarddb.com/sim/sim-studio-v0-5-95-adds-gemini-3-1-pro-support-and-4-new-integrations--mmxnddmx
- Original: https://github.com/simstudioai/sim/releases/tag/v0.5.95

## New Features

**Model Support**: Added support for Google's Gemini 3.1 Pro Preview model, with updated thinking levels for the existing Gemini 3 Pro model, enabling developers to leverage the latest LLM capabilities.

**Integrations**: Four new integrations now available—**Redis**, **Upstash**, **Algolia**, and **RevenueCat**—expanding the toolkit for data management, search, and monetization workflows.

**Tables Feature**: Introduced a new tables feature for managing structured data within workflows.

**Enhanced Tool Support**: Advanced fields added for YouTube and Vercel integrations. New integrations for Cloudflare and Microsoft Dataverse bring enterprise infrastructure connectivity to the platform.

## Improvements & Bug Fixes

- Improved resolver behavior to prevent empty sentinels from passing invalid references to text inputs
- Fixed corrupted sticky disk cache issue on Blacksmith infrastructure
- Updated Node.js version alignment in trigger.dev container builds
- Resolved concurrent workflow snapshot conflicts by switching from insert to upsert operations
- Enhanced audit logging with lazy actor name/email resolution
- Enforced unique workflow names at the folder level to prevent naming conflicts
- Fixed Jira Service Management block to require serviceDeskId parameter

---

### v0.16.3
- Published: 2026-02-20T21:25:03+00:00
- Products: 
- Tags: release, feature, integration
- URL: https://standarddb.com/ollama/ollama-v0-16-3-adds-cline-cli-integration-and-support-for-gemma-3-llama-3-qwen-3--xsztvgzs
- Original: https://github.com/ollama/ollama/releases/tag/v0.16.3

## What's New

**Cline CLI Integration**: A new `ollama launch cline` command has been added, enabling direct integration with the Cline CLI for streamlined workflow.

**Improved Model Picker**: The `ollama launch <integration>` command now always displays the model picker, ensuring a consistent user experience across all integration points.

**Expanded Architecture Support**: Added support for three major model architectures to the MLX runner:
- Gemma 3
- Llama 3
- Qwen 3

This expansion enables better compatibility with the latest generation of open-source models and improves performance on Apple Silicon devices.

## For Developers

Users running Ollama can immediately take advantage of the expanded model support when using the MLX runner. Those using Cline can now launch Ollama directly from the CLI. Update to v0.16.3 to access these improvements.

---

### Organization-level Copilot usage metrics dashboard available in public preview
- Published: 2026-02-20T19:01:39+00:00
- Products: GitHub Copilot, GitHub Dependabot
- Tags: feature, platform, api
- URL: https://standarddb.com/github/github-copilot-organization-level-usage-metrics-dashboard-enters-public-preview--hqdbbnqr
- Original: https://github.blog/changelog/2026-02-20-organization-level-copilot-usage-metrics-dashboard-available-in-public-preview

## Organization-Level Analytics Now Available

GitHub has expanded its Copilot usage metrics functionality to the organization level, moving beyond the previously enterprise-only dashboards. Organization owners can now view detailed Copilot usage metrics directly in the GitHub UI, giving teams better visibility into adoption patterns and trends without needing to rely on enterprise-level reporting.

## Key Capabilities

The dashboard provides organization-scoped insights that mirror the recently released organization usage APIs, presenting data in an easy-to-scan format. Organization owners and users with appropriate custom roles can access the dashboard (requiring the `View Organization Copilot Metrics` permission), as long as Copilot usage metrics are enabled.

**Key benefits include:**
- Track Copilot adoption and usage trends at the organization level
- Grant access using least-privilege organization-scoped roles without exposing enterprise-wide metrics
- Reduce dependency on enterprise reporting for day-to-day organizational insights

## Availability and Access

All organization types can use this feature, including:
- Enterprise-owned organizations
- Standalone Free or Team organizations

**Important consideration:** Organization-level totals will not match enterprise totals when organizations belong to an enterprise. Since users can be members of multiple organizations, their usage may appear across multiple organization reports, while enterprise reporting deduplicates users across organizations.

Developers can reference the [Copilot metrics documentation](https://docs.github.com/copilot/concepts/copilot-usage-metrics/copilot-metrics) for a complete list of available metrics.

---

### Code Mode: give agents an entire API in 1,000 tokens
- Published: 2026-02-20T14:00:00+00:00
- Products: Cloudflare Workers, Cloudflare
- Tags: integration, api, feature, sdk, open-source
- URL: https://standarddb.com/cloudflare/cloudflare-launches-mcp-server-using-code-mode-reducing-api-context--byqjpkxp
- Original: https://blog.cloudflare.com/code-mode-mcp

## Code Mode: Efficient API Access for AI Agents

Cloudflare announced a new MCP server that gives AI agents access to the entire Cloudflare API while consuming only around 1,000 tokens—a 99.9% reduction compared to traditional MCP implementations that would require 1.17 million tokens.

### How Code Mode Works

Instead of exposing thousands of individual API tools, the server uses **Code Mode**—a technique that lets models write and execute code against a typed SDK. The MCP server exposes just two tools:

- **`search()`** — Allows agents to query the Cloudflare OpenAPI spec by writing JavaScript code to filter endpoints by product, path, tags, or metadata
- **`execute()`** — Enables agents to write JavaScript code that makes Cloudflare API requests, handles pagination, chains operations, and returns results

Both tools run generated code inside a **Dynamic Worker isolate**—a lightweight V8 sandbox with no file system access, no environment variable leakage, and disabled external fetches by default.

### Key Benefits

- **Massive context savings**: Provides access to hundreds of API endpoints without inflating model context windows
- **Type-safe SDK**: Agents write code against a fully typed representation of the API, improving reliability
- **Composable operations**: Agents can chain multiple API calls and handle complex logic in a single execution
- **Security by design**: Code runs in an isolated sandbox with explicit fetch controls

### Getting Started

The Cloudflare MCP server is available today and can be used with any MCP-compatible AI agent. Cloudflare is also **open-sourcing the Code Mode SDK** in the [Cloudflare Agents SDK](https://github.com/cloudflare/agents), enabling developers to apply the same pattern to their own MCP servers and AI agents.

The approach addresses a fundamental tension in agent design: agents need many tools to do useful work, but each tool consumes precious context window space. Code Mode solves this by letting the model itself determine which operations to perform through executable code rather than pre-defining tool boundaries.

---

### Making frontier cybersecurity capabilities available to defenders
- Published: 2026-02-20T00:00:00+00:00
- Products: Claude Code, Claude
- Tags: feature, security, api, release
- URL: https://standarddb.com/anthropic/anthropic-launches-claude-code-security-in-limited-preview-found-500-zero-day--pdpxnvwt
- Original: https://www.anthropic.com/news/claude-code-security

## Claude Code Security: AI-Powered Vulnerability Detection

Anthropic announced **Claude Code Security**, a new capability integrated into Claude Code on the web designed to help security teams automatically discover and patch vulnerabilities in source code. Unlike traditional static analysis tools that match code against known vulnerability patterns, Claude Code Security reasons about code semantics the way human security researchers do—understanding component interactions, tracing data flow, and identifying complex vulnerabilities like business logic flaws and broken access control.

## Key Technical Capabilities

The tool includes several safeguards and verification mechanisms:

- **Intelligent analysis**: Claude reads code holistically to detect context-dependent vulnerabilities that rule-based tools typically miss
- **Multi-stage verification**: Every finding undergoes re-examination to filter out false positives before reaching analysts
- **Severity and confidence ratings**: Findings are prioritized and flagged with confidence levels to guide remediation efforts
- **Human-in-the-loop approval**: All identified issues and suggested patches require developer review and approval—Claude identifies problems and suggests solutions, but developers retain full control

## Proven Track Record

Claude Code Security builds on over a year of cybersecurity research. Using Claude Opus 4.6, Anthropic's Frontier Red Team discovered **over 500 zero-day vulnerabilities** in production open-source codebases—critical bugs that had gone undetected for years despite expert review. The team has also:

- Participated in competitive Capture-the-Flag cybersecurity events
- Partnered with Pacific Northwest National Laboratory on critical infrastructure defense research
- Validated the tool's effectiveness by using it to secure Anthropic's own systems

## Availability and Access

The limited research preview is now available to Enterprise and Team customers, with expedited access provided for open-source maintainers. Organizations can apply for access at [claude.com/contact-sales/security](https://claude.com/contact-sales/security). Since the tool is built on Claude Code, security teams can review findings and iterate on fixes within tools they already use.

Anthropic emphasizes this is a pivotal moment for cybersecurity—as AI tools become more effective at finding vulnerabilities, both defenders and attackers will leverage these capabilities. The goal is to ensure defenders can identify and patch security issues faster than attackers can exploit them.

---

### Cloudflare Tunnel - Manage Cloudflare Tunnel directly from the main Cloudflare Dashboard
- Published: 2026-02-20T00:00:00+00:00
- Products: Cloudflare Workers, Cloudflare
- Tags: feature, platform, integration
- URL: https://standarddb.com/cloudflare/cloudflare-tunnel-now-available-in-main-dashboard-with-full-lifecycle-management--sjfrrzvy
- Original: https://developers.cloudflare.com/changelog/2026-02-20-tunnel-core-dashboard

## Tunnel Management in Core Dashboard

Cloudflare has moved Tunnel management to the main Cloudflare Dashboard at **Networking > Tunnels**, giving developers a unified interface for securing origin servers and public applications. This consolidation reduces friction by eliminating the need to context-switch between the core dashboard and Cloudflare One dashboard for different tunnel use cases.

## Key Features

The new dashboard experience includes:

- **Full lifecycle management**: Create, configure, delete, and monitor tunnels from a single location
- **Native integrations**: Reference tunnels by name when configuring DNS records and Workers VPC connections—no more UUID copy-pasting
- **Real-time monitoring**: Track replica status and tunnel health directly in the dashboard
- **Unified routing interface**: Manage all ingress routes (public applications, private hostnames, CIDRs, and Workers VPC services) from one interactive map

## Dashboard Organization

Cloudflare maintains two complementary dashboards for different workflows:

- **Core Dashboard** (`Networking > Tunnels`): For securing origin servers and public applications with CDN, WAF, and DDoS protection; also for connecting Workers to private services via Workers VPC
- **Cloudflare One Dashboard** (`Zero Trust > Networks > Connectors`): For Zero Trust access policies, private application routing, and private mesh networking

Both dashboards offer complete tunnel management—developers should choose based on their primary use case and workflow.

## Getting Started

Developers new to Cloudflare Tunnel can follow the [get started guide](https://developers.cloudflare.com/cloudflare-one/networks/connectors/cloudflare-tunnel/get-started/create-remote-tunnel/) or explore advanced use cases like SSH server tunneling and Kubernetes deployments.

---

### DDoS protection, Railway domains, better canvas state
- Published: 2026-02-20T00:00:00+00:00
- Products: 
- Tags: release, feature, security, integration
- URL: https://standarddb.com/railway/railway-adds-default-ddos-protection-via-fastly-launches-domain-registration--zyqmzyxc
- Original: https://railway.com/changelog/2026-02-20-domains

## DDoS Protection by Default

Railway has partnered with **Fastly** to deliver built-in DDoS protection across all public services. All traffic now routes through Fastly's global network of 100+ points of presence, where malicious traffic is filtered before reaching your infrastructure. This protection is **enabled automatically with no configuration required**—no proxy setup, no feature flags, no plan upgrades needed.

Previously, customers had to manage DDoS protection separately through external providers, requiring context-switching between platforms and manual DNS configuration. This integration eliminates that friction entirely.

## Domain Registration and Management

Railway is introducing integrated **domain registration and management** for Priority Boarding tier customers. You can now search, purchase, and manage domains directly within the Railway dashboard—supporting over 250 TLDs with AI-powered naming suggestions.

Key features include:
- **Automatic DNS configuration** when purchasing a domain from a service's networking settings
- **WHOIS privacy and auto-renewal enabled by default** on all domains
- **Cost-based pricing**, rounded to the nearest dollar
- **Centralized management** at `railway.com/workspace/domains` for expiry dates, attached services, and payment status
- **Import/export layouts** between environments for flexible organization

No more juggling external registrars and DNS providers—everything happens within Railway.

## Per-Environment Canvas Layouts

Canvas state (node positions, groupings, edge anchors) is now **scoped per environment** instead of shared across the entire project. Previously, all team members and environments shared one layout, causing phantom nodes to appear, layout conflicts when multiple people edited simultaneously, and misaligned positions when deploying templates across environments.

Now, creating a new environment copies the layout as a starting point, but the two remain independent. This allows teams to organize production differently from staging, arrange by dependency chains in one environment and by team ownership in another, and avoid layout conflicts during concurrent edits.

## Additional Improvements

Minor fixes and enhancements include: variable autocompletion and markdown support in template deployments, centered canvas views when side panels are open, improved GitHub installation error visibility, corrected domain TXT record formatting, and the ability to paste GitHub URLs directly in the command palette.

---

### Cloudflare Tunnel - Manage Cloudflare Tunnel directly from the main Cloudflare Dashboard
- Published: 2026-02-20T00:00:00+00:00
- Products: Cloudflare Workers, Cloudflare
- Tags: feature, platform, integration
- URL: https://standarddb.com/cloudflare/cloudflare-tunnel-management-now-available-in-core-dashboard--dhqzqjdl
- Original: https://developers.cloudflare.com/changelog/post/2026-02-20-tunnel-core-dashboard

## Tunnel Management Now in Core Dashboard

Cloudflare has brought Tunnel management to its main Dashboard, moving beyond the Cloudflare One platform. Developers can now access the full Tunnel experience at **Networking > Tunnels** in the core Dashboard, streamlining workflows for those securing origin servers and public applications.

## Key Capabilities

The new dashboard experience includes:

- **Full lifecycle management**: Create, configure, delete, and monitor Tunnels from a single interface
- **Native integrations**: Select Tunnels by name when setting up DNS records and Workers VPC connections—eliminating manual UUID copying
- **Real-time monitoring**: Track replica health status and Tunnel availability directly in the dashboard
- **Unified routing map**: Manage all ingress routes (public applications, private hostnames, private CIDRs, and Workers VPC services) from an interactive interface

## Dashboard Selection Guide

**Use the Core Dashboard** (Networking > Tunnels) for:
- Securing origin servers and public applications with CDN, WAF, Load Balancing, and DDoS protection
- Connecting Workers to private services via Workers VPC

**Use the Cloudflare One Dashboard** (Zero Trust > Networks > Connectors) for:
- Implementing Zero Trust access policies on public applications
- Managing private application access
- Building private mesh networks

Both dashboards provide complete Tunnel functionality—choose based on your primary use case and workflow.

---

### Agents, Workers - @cloudflare/codemode v0.1.0: a new runtime agnostic modular architecture
- Published: 2026-02-20T00:00:00+00:00
- Products: Cloudflare Workers
- Tags: release, sdk, api, breaking-change, feature
- URL: https://standarddb.com/cloudflare/cloudflare-releases-cloudflare-codemode-v0-1-0-with-runtime-agnostic-sdk--qkxwrppt
- Original: https://developers.cloudflare.com/changelog/post/2026-02-20-codemode-sdk-rewrite

## Runtime-Agnostic Architecture

The `@cloudflare/codemode` package has been completely rewritten to support a modular, runtime-agnostic design. Code Mode continues to enable LLMs to write and execute code that orchestrates your tools, rather than calling them sequentially, resulting in significant token savings and improved model performance on complex tasks.

## Key Improvements

**New `Executor` Interface**
- Minimal contract for code execution: `execute(code, fns)`
- Implement your own executor for any code sandboxing primitive or runtime
- Decouples the SDK from specific runtime dependencies

**`createCodeTool()` Function**
- Returns a standard AI SDK `Tool` for use in AI agents
- New import path: `@cloudflare/codemode/ai`
- Replaces the previous `experimental_codemode()` API

**`DynamicWorkerExecutor`**
- Runs generated code in [Dynamic Workers](https://developers.cloudflare.com/workers/runtime-apis/bindings/worker-loader/)
- Network isolation: `fetch()` and `connect()` blocked by default
- Console capture: Logs from `console.log/warn/error` returned in `ExecuteResult.logs`
- Configurable execution timeout (default 30 seconds)

## Migration Guide

This release includes **breaking changes**:
- `experimental_codemode()` and `CodeModeProxy` have been removed
- The package no longer owns LLM calls or model selection
- Update imports to use `createCodeTool()` from `@cloudflare/codemode/ai`

To upgrade: `npm i @cloudflare/codemode@latest`

Refer to the [Code Mode documentation](https://developers.cloudflare.com/agents/api-reference/codemode/) for complete API reference and additional examples.

---

### CASB - Understand CASB findings instantly with Cloudy Summaries
- Published: 2026-02-20T00:00:00+00:00
- Products: Cloudflare
- Tags: feature, integration, platform
- URL: https://standarddb.com/cloudflare/cloudflare-casb-now-includes-ai-powered-cloudy-summaries-for-instant-security--rxfdydkk
- Original: https://developers.cloudflare.com/changelog/post/2026-02-20-cloudy-in-casb

## AI-Powered Security Summaries Now Available

Cloudflare has launched **Cloudy Summaries** for its Cloud Access Security Broker (CASB) product, bringing AI-assisted analysis directly into security findings. This feature automatically generates plain-language summaries of complex security issues detected across your SaaS environment.

### What's New

The feature integrates Cloudflare's Cloudy AI into CASB Posture Findings to help security teams and IT administrators quickly understand:

- **What was detected** — Detailed explanations of misconfigurations, risky third-party app permissions, and data exposures
- **Why it matters** — Clear assessment of potential security impact, including data breach risks and unauthorized access vulnerabilities
- **How to fix it** — Actionable recommendations for remediation and environment hardening

### Supported Integrations

Cloudy Summaries are available across all CASB integrations, including:
- Microsoft 365
- Google Workspace
- Salesforce
- GitHub
- AWS
- Slack
- Dropbox

### How to Use

Simply navigate to **Cloud and SaaS findings** in the Cloudflare One dashboard, open any finding's details, and review the AI-generated summary. Users can provide feedback via thumbs-up/thumbs-down ratings to improve future summarization accuracy.

The feature is available to all Cloudflare CASB users immediately.

---

### Agents, Workers - @cloudflare/codemode v0.1.0: a new runtime agnostic modular architecture
- Published: 2026-02-20T00:00:00+00:00
- Products: Cloudflare Workers
- Tags: release, sdk, feature, breaking-change
- URL: https://standarddb.com/cloudflare/cloudflare-releases-cloudflare-codemode-v0-1-0-with-runtime-agnostic--pjlmhwtq
- Original: https://developers.cloudflare.com/changelog/2026-02-20-codemode-sdk-rewrite

## @cloudflare/codemode v0.1.0 Released

Cloudflare has released a major rewrite of the `@cloudflare/codemode` package, transitioning it to a **modular, runtime-agnostic SDK** designed to power agent workflows. Code Mode enables LLMs to write and execute code that orchestrates multiple tools at once, rather than calling them individually—yielding significant token savings, reduced context window pressure, and improved model performance.

### Key Breaking Changes

The update introduces several breaking changes developers should be aware of:

- **Removed `experimental_codemode()` and `CodeModeProxy`** — The SDK no longer owns LLM calls or model selection, allowing greater flexibility in how you integrate it with different AI frameworks.
- **New import path** — `createCodeTool()` is now exported from `@cloudflare/codemode/ai` instead of the root package.

### New Features

**`createCodeTool()`** returns a standard AI SDK `Tool` compatible with popular AI frameworks like the Vercel AI SDK. This enables seamless integration into your existing agent architectures.

**`Executor` interface** provides a minimal contract with a single `execute(code, fns)` method, making it easy to implement custom code execution sandboxes or integrate with any runtime environment.

### DynamicWorkerExecutor

The bundled `DynamicWorkerExecutor` runs code in Cloudflare's Dynamic Worker environment with production-ready features:

- **Network isolation** — `fetch()` and `connect()` are blocked by default (`globalOutbound: null`) for secure code execution
- **Console capture** — Logs from `console.log()`, `warn()`, and `error()` are captured and returned in execution results
- **Configurable timeout** — Execution limits default to 30 seconds but are customizable per invocation

### Getting Started

To upgrade, install the latest version:

```bash
npm i @cloudflare/codemode@latest
```

Configure your Wrangler project to include a worker loader binding:

```toml
[[worker_loaders]]
binding = "LOADER"
```

Then create a code tool and integrate it into your AI agent using the Vercel AI SDK or similar framework. See the [Code Mode documentation](https://developers.cloudflare.com/agents/api-reference/codemode/) for full API reference and additional examples.

---

### GitHub Copilot support in Zed generally available
- Published: 2026-02-19T22:24:53+00:00
- Products: GitHub Copilot
- Tags: feature, integration, platform
- URL: https://standarddb.com/github/github-copilot-support-in-zed-editor-now-generally-available--nnplhccy
- Original: https://github.blog/changelog/2026-02-19-github-copilot-support-in-zed-generally-available

## General Availability

GitHub has announced official support for GitHub Copilot in Zed, the high-performance multiplayer code editor built in Rust by the creators of Atom and Tree-sitter. This partnership allows developers to use their existing Copilot subscriptions across additional development tools and workflows.

## Supported Subscriptions

All developers with **paid GitHub Copilot subscriptions** can now authenticate into Zed:
- Copilot Pro
- Copilot Pro+
- Copilot Business
- Copilot Enterprise

No additional AI license or subscription is required beyond your existing Copilot subscription.

## Getting Started

Using Copilot in Zed is straightforward:
1. Open **Settings** in Zed
2. Navigate to the **GitHub Copilot Chat** section
3. Click **Sign in to use GitHub Copilot**
4. Follow the authentication flow in the modal
5. Start using Copilot-powered features in your editor

## About Zed

Zed is a next-generation code editor emphasizing performance and collaborative development, built in Rust by the team behind Atom (deprecated) and Tree-sitter. The editor is designed for modern development workflows with native multiplayer capabilities.

For more details, visit [Zed's Copilot documentation](https://zed.dev/docs/ai/llm-providers#github-copilot-chat).

---

### Copilot coding agent model picker for Copilot Business and Enterprise
- Published: 2026-02-19T19:55:01+00:00
- Products: GitHub Copilot
- Tags: feature, release
- URL: https://standarddb.com/github/github-copilot-coding-agent-gains-model-picker-for-business-and-enterprise-users--vcfkknlb
- Original: https://github.blog/changelog/2026-02-19-model-picker-for-copilot-coding-agent-for-copilot-business-and-enterprise-users

## Model Selection Now Available for Business and Enterprise

GitHub has extended model picker functionality to **Copilot Business and Copilot Enterprise users**, building on the model selection feature [launched in December for Copilot Pro and Pro+ subscribers](https://github.blog/changelog/2025-12-08-model-picker-for-copilot-coding-agent-for-copilot-pro-and-pro-subscribers/). This feature allows teams to choose the optimal AI model for each coding task assigned to Copilot's autonomous agent.

## Available Models

When assigning an issue to Copilot on github.com or starting a task from the agents tab, agents panel, GitHub Mobile, or Raycast launcher, you can select from:

- **Claude models**: Opus 4.5, Opus 4.6, Sonnet 4.5, Sonnet 4.6
- **OpenAI GPT models**: GPT-5.1-Codex-Max, GPT-5.2-Codex, GPT-5.3-Codex
- **Auto** (default): Copilot automatically optimizes speed and performance by selecting the best available model

## Important Notes for Administrators

If your organization's administrator has not enabled any of the models above, the model picker will not appear in the UI, and Claude Sonnet 4.6 will be used as the default model. Organizations can configure which models are available to their teams based on their specific needs and preferences.

For detailed instructions on model selection and configuration, see the [Copilot coding agent model selection documentation](https://docs.github.com/copilot/how-tos/use-copilot-agents/coding-agent/changing-the-ai-model).

---

### Gemini 3.1 Pro is now in public preview in GitHub Copilot
- Published: 2026-02-19T17:30:18+00:00
- Products: GitHub Copilot
- Tags: release, model, feature
- URL: https://standarddb.com/github/gemini-3-1-pro-now-available-in-github-copilot-with-improved-tool-precision-and--cpgmwycp
- Original: https://github.blog/changelog/2026-02-19-gemini-3-1-pro-is-now-in-public-preview-in-github-copilot

## Gemini 3.1 Pro Launches in GitHub Copilot

Google's Gemini 3.1 Pro, designed specifically for agentic coding workflows, is now available in public preview within GitHub Copilot. Early testing shows the model excels at creating efficient edit-then-test loops with notably high tool precision, allowing developers to resolve issues with fewer tool calls compared to benchmarks.

### Availability & Access

Gemini 3.1 Pro is available to Copilot Pro, Pro+, Business, and Enterprise users. You can select this model through the model picker in:

- **Visual Studio Code**: Chat, ask, edit, and agent modes
- **Visual Studio**: Agent and ask modes
- **github.com**: Web interface
- **GitHub Mobile**: iOS and Android apps

The rollout is gradual, so availability may expand over the coming weeks.

### Admin Requirements

Copilot Enterprise and Copilot Business administrators need to enable the Gemini 3.1 Pro policy in Copilot settings to grant their organization access.

### Getting Started

For a complete list of available models and to get started using Copilot, review the [official documentation on supported models](https://docs.github.com/copilot/reference/ai-models/supported-models). You can also join the GitHub Community to share feedback and discuss your experience with the new model.

---

### Gemini 3.1 Pro: A smarter model for your most complex tasks
- Published: 2026-02-19T16:06:14+00:00
- Products: Gemini
- Tags: release, model, feature, api
- URL: https://standarddb.com/google/google-releases-gemini-3-1-pro-with-77-1-arc-agi-score-doubling-reasoning--djctvlcp
- Original: https://deepmind.google/blog/gemini-3-1-pro-a-smarter-model-for-your-most-complex-tasks

## Gemini 3.1 Pro Available Now

Google has released **Gemini 3.1 Pro**, the next generation of its flagship AI model, rolling out across consumer and developer products. This upgraded core intelligence represents a major step forward in reasoning capabilities and is now available through multiple platforms.

### Platform Availability

- **Developers**: Available in preview via Google AI Studio, Gemini CLI, Google Antigravity, and Android Studio
- **Enterprises**: Accessible through Vertex AI and Gemini Enterprise
- **Consumers**: Rolling out via the Gemini app and NotebookLM

### Performance Improvements

3.1 Pro delivers substantial improvements in complex reasoning:

- **ARC-AGI-2 benchmark**: Achieved a verified score of **77.1%**, more than **double the reasoning performance** of Gemini 3 Pro
- Designed as a smarter, more capable baseline for complex problem-solving tasks
- Particularly strong at solving entirely new logic patterns and advancing agentic workflows

### Practical Applications

The improved reasoning capabilities enable developers and users to tackle advanced tasks including:

- **Visual explanations** of complex topics with clear, interactive visuals
- **Data synthesis** combining multiple sources into unified views
- **Advanced code generation**, including animated SVGs from text prompts
- **Complex system integration**, such as building live dashboards from multiple APIs
- **Interactive design and prototyping**, enabling sensory-rich interfaces with hand-tracking and generative elements

### What Developers Need to Know

3.1 Pro is shipping in preview as the core upgrade underlying Gemini 3 Deep Think. This is a significant reasoning improvement suitable for applications requiring sophisticated problem-solving and multi-step reasoning tasks. Developers can begin testing the model immediately through the listed platforms.

---

### Grok Imagine Video on AI Gateway
- Published: 2026-02-19T13:00:00+00:00
- Products: Vercel AI SDK, Vercel
- Tags: feature, integration, api, model
- URL: https://standarddb.com/vercel/vercel-ai-gateway-adds-grok-imagine-video-for-text-to-video-and-video-editing--jdvxtmgz
- Original: https://vercel.com/changelog/grok-imagine-video-on-ai-gateway

## Grok Imagine Video Now Available

Vercel has integrated xAI's Grok Imagine Video into its AI Gateway, providing developers with a powerful video generation tool that produces high-quality clips with realistic motion and natural audio. The model is now available for Pro and Enterprise plan subscribers, as well as paid AI Gateway users.

## Key Capabilities

**Video Generation Features**:
- **Text-to-video**: Generate videos from text descriptions with configurable duration and resolution (up to 1280x720)
- **Video editing**: Transform existing videos by changing style, swapping objects, or altering scenes
- **Audio generation**: Native audio synthesis with expressive voices and accurate lip-sync
- **Fast generation**: Creates clips in seconds rather than minutes
- **Instruction following**: Handles complex prompts and follow-up instructions for scene tweaking

## Getting Started

Developers can access Grok Imagine Video through three methods:

1. **v0 Grok Creative Studio**: A template-based UI for creating and showcasing video generations without code
2. **AI SDK 6**: Use the `generateVideo()` function to integrate video generation programmatically with full control over model parameters
3. **AI Gateway Playground**: Experiment with no-code tools, compare providers, tweak prompts, and download results directly in the browser

## Implementation Example

The AI SDK integration is straightforward:
```
const { videos } = await generateVideo({
  model: 'xai/grok-imagine-video',
  prompt: 'Your text description',
  aspectRatio: '16:9',
  duration: 3,
});
```

Advanced features like reference-to-video transformations enable style transfer and scene modifications by passing video URLs and transformation prompts.

---

### Veo video models on AI Gateway
- Published: 2026-02-19T13:00:00+00:00
- Products: Vercel AI SDK, Vercel
- Tags: feature, integration, api, model
- URL: https://standarddb.com/vercel/vercel-adds-google-veo-video-models-to-ai-gateway-with-native-audio-synthesis--sdwnznyv
- Original: https://vercel.com/changelog/veo-video-models-on-ai-gateway

## Veo Video Models Now Available

Google's Veo video generation models are now integrated into Vercel's AI Gateway, bringing photorealistic video generation capabilities directly to developers. The models support multiple workflows including text-to-video generation, image-to-video animation, and native audio synthesis that automatically generates sound effects, ambient audio, and dialogue matched to the video content.

## Key Capabilities

The available Veo models deliver:

- **Up to 1080p Resolution:** Generate videos at 720p and 1080p quality
- **Native Audio Generation:** Automatically create realistic sound effects and ambient audio that synchronizes with video content
- **Photorealistic Quality:** Specialized for nature, wildlife, and cinematic scenes
- **Image-to-Video:** Animate still photos with natural motion
- **Fast Mode:** Quicker generation iterations available for rapid prototyping

## Getting Started

Two integration paths are available:

**AI SDK 6**: Generate videos programmatically using the `generateVideo` function with model selection and configuration options including aspect ratio, duration, and audio generation settings.

**AI Gateway Playground**: A no-code interface for experimenting with video models, comparing providers, tweaking prompts, and downloading results directly in the browser.

## Available Models

Four Veo model variants are currently accessible:
- `google/veo-3.1-generate-001` (latest, highest quality)
- `google/veo-3.1-fast-generate-001` (fast iterations)
- `google/veo-3.0-generate-001` (full quality)
- `google/veo-3.0-fast-generate-001` (fast mode)

## Access and Availability

Video generation is currently in beta and available for Pro and Enterprise plans as well as paid AI Gateway users. The feature is fully documented with video generation quick starts and detailed configuration guidance in the Vercel documentation.

---

### Kling video models on AI Gateway
- Published: 2026-02-19T13:00:00+00:00
- Products: Vercel AI SDK, Vercel
- Tags: feature, api, integration, model
- URL: https://standarddb.com/vercel/vercel-ai-gateway-adds-kling-video-models-with-multishot-and-audio-generation--wvyrmqfl
- Original: https://vercel.com/changelog/kling-video-models-on-ai-gateway

## Kling Video Models Now Available

Kling video models are now integrated into Vercel's AI Gateway, making state-of-the-art video generation accessible through a unified API. This expansion includes the latest Kling 3.0 models alongside earlier versions, giving developers multiple options for performance and capability trade-offs.

### Key Capabilities

Kling models are optimized for several video generation scenarios:

- **Image-to-Video**: Strong animation of still images into video clips with realistic motion and physics
- **Multishot Narratives**: Kling 3.0 can generate multi-scene videos from a single narrative prompt with intelligent shot composition
- **High-Quality Output**: Supports up to 1080p generation in pro mode
- **Audio Generation**: Creates synchronized sound effects and ambient audio alongside video
- **Precise Control**: First and last frame control allows developers to specify exact start and end frames for scene transitions

### Getting Started

Video generation is available for Pro and Enterprise plan users and paid AI Gateway customers. Developers have two main integration paths:

- **AI SDK 6**: Use the `generateVideo` method for programmatic video generation with full control over model parameters and provider-specific options
- **AI Gateway Playground**: A no-code interface for experimenting with video models, comparing providers, tweaking prompts, and downloading results directly from the browser

### Available Models

The service offers six Kling model variants across different generations and use cases:

| Model | Type | Highlights |
|-------|------|-----------|
| Kling v3.0 Text-to-Video | Latest generation | Multishot support, highest quality |
| Kling v3.0 Image-to-Video | Latest generation | Multiple frame control with v3 quality |
| Kling v2.6 Text-to-Video | Production-ready | Audio generation support |
| Kling v2.6 Image-to-Video | Production-ready | Image reference with audio |
| Kling v2.5 Turbo Text-to-Video | Speed-optimized | Faster generation times |
| Kling v2.5 Turbo Image-to-Video | Speed-optimized | Faster generation for images |

### Usage Examples

The AI SDK provides straightforward integration for common workflows. Simple text-to-video generation requires only a prompt and basic parameters like aspect ratio and duration. Advanced multishot narratives can be created by structuring prompts as multi-scene stories and enabling Kling-specific options like intelligent shot composition and synchronized audio. For precise scene transitions, developers can provide both start and end frame images to control exactly how videos begin and conclude.

---

### Streamdown 2.3 — Refreshed design and interactive playground
- Published: 2026-02-19T13:00:00+00:00
- Products: Vercel
- Tags: release, feature
- URL: https://standarddb.com/vercel/streamdown-2-3-adds-interactive-playground-and-unified-design-system-for--tcrhpltl
- Original: https://vercel.com/changelog/streamdown-2-3

## Design & UX Improvements

Streamdown 2.3 standardizes the visual experience with a unified nested-card design applied across tables, code blocks, and Mermaid diagrams. Action buttons now remain sticky during scroll for easier access, and code blocks render plain text immediately—reducing perceived latency while syntax highlighting loads asynchronously.

## Interactive Playground

A new interactive playground accelerates development and testing workflows. Developers can now execute custom markdown in real-time, edit configuration props directly, and see results instantly—eliminating the need to set up a local project for rapid experimentation.

## Enhanced Rendering APIs

Several new hooks and utilities improve rendering control:
- **`useIsCodeFenceIncomplete` hook**: Detects in-progress fenced code blocks during streaming, essential for streaming-based markdown rendering
- **Markdown copying for tables**: Tables now support exporting content as formatted Markdown
- **HTML indentation normalization**: New property handles inconsistent whitespace in raw HTML input
- **Improved image error handling**: Custom error messaging for failed image renders

## Documentation & Quality

Documentation has been reorganized with dedicated pages for CJK, Math, and Mermaid plugins, and the homepage now directly links to templates for faster onboarding. This release clears the bug backlog, addressing issues with nested HTML block parsing, custom tag handling, Mermaid diagram artifacts, and Shiki syntax engine inconsistencies.

---

### Gemini 3.1 Pro is live on AI Gateway
- Published: 2026-02-19T13:00:00+00:00
- Products: Vercel AI SDK, Vercel
- Tags: release, model, api, integration, feature
- URL: https://standarddb.com/vercel/vercel-ai-gateway-now-supports-google-s-gemini-3-1-pro-with-medium-thinking--rdzvvkpq
- Original: https://vercel.com/changelog/gemini-3-1-pro-is-live-on-ai-gateway

## Gemini 3.1 Pro Now Available on AI Gateway

Google's Gemini 3.1 Pro Preview is now live on Vercel's AI Gateway, giving developers access to Google's latest model through a unified API. This release brings quality improvements specifically optimized for software engineering tasks and agentic workflows, with enhanced capabilities for real-world applications in finance and spreadsheet operations.

### Key Features

The new model introduces **medium thinking level support**, allowing developers to fine-tune the trade-off between cost, performance, and inference speed. This gives fine-grained control over how the model allocates computational resources for different tasks.

- **Improved efficiency**: More efficient thinking across use cases with reduced token consumption
- **Better performance**: Quality improvements while maintaining or reducing costs
- **Specialized strengths**: Enhanced for software engineering, agentic workflows, finance, and spreadsheet applications

### How to Use

To access Gemini 3.1 Pro, set the model identifier to `google/gemini-3.1-pro-preview` in the Vercel AI SDK and configure the thinking level via provider options:

```tsx
const result = streamText({
  model: 'google/gemini-3.1-pro-preview',
  prompt: 'Your task here',
  providerOptions: {
    google: {
      thinking_level: 'medium',
    },
  },
});
```

### Why AI Gateway

Vercel's AI Gateway provides unified access across multiple model providers with built-in observability, usage tracking, cost management, automatic retries, and intelligent failover—all without exposing provider API keys directly.

Developers can explore the model on the [AI Gateway leaderboard](https://vercel.com/ai-gateway/leaderboards) or test it in the [model playground](https://vercel.com/ai-gateway/models/gemini-3.1-pro-preview) before deploying.

---

### Wan models on AI Gateway
- Published: 2026-02-19T13:00:00+00:00
- Products: Vercel AI SDK, Vercel
- Tags: feature, api, integration, model
- URL: https://standarddb.com/vercel/vercel-adds-alibaba-s-wan-video-generation-models-to-ai-gateway--kbhrphbc
- Original: https://vercel.com/changelog/wan-models-on-ai-gateway

## Video Generation with Wan Models Now Available

Vercel has integrated Alibaba's Wan video generation models into AI Gateway, enabling developers to generate stylized videos with smooth motion and consistent character appearance. The integration brings multiple video generation capabilities directly into the AI SDK and web-based playground.

## Supported Video Generation Types

The Wan model suite includes:

- **Text-to-Video (`alibaba/wan-v2.6-t2v`)**: Generate videos from text prompts with support for detailed styling instructions
- **Image-to-Video (`alibaba/wan-v2.6-i2v`)**: Animate still images into videos
- **Image-to-Video Flash (`alibaba/wan-v2.6-i2v-flash`)**: Faster generation for quick iterations
- **Reference-to-Video (`alibaba/wan-v2.6-r2v`)**: Extract character appearance and voice from reference images/videos to generate new scenes
- **Reference-to-Video Flash (`alibaba/wan-v2.6-r2v-flash`)**: Expedited character transfer

All models support flexible resolutions at 480p, 720p, and 1080p with configurable duration.

## Getting Started

Developers can access Wan models through two interfaces:

1. **AI SDK 6**: Use the `generateVideo()` function to programmatically generate videos with full configuration control
2. **AI Gateway Playground**: Experiment with video models visually without code, compare providers, and download results

## Availability and Access

Video generation is currently in beta and available to Pro and Enterprise plan subscribers, as well as paid AI Gateway users. The models support detailed prompts for style specification (e.g., anime, watercolor, neon-lit scenes) and character referencing for multi-subject scenes.

Developers can access the full [video generation documentation](https://vercel.com/docs/ai-gateway/capabilities/video-generation) for advanced configuration options and quickstart scripts.

---

### Private storage for Vercel Blob, now available in public beta
- Published: 2026-02-19T13:00:00+00:00
- Products: Vercel
- Tags: feature, release, sdk, api
- URL: https://standarddb.com/vercel/vercel-blob-adds-private-storage-in-public-beta-for-authenticated-file-access--zxvklrkq
- Original: https://vercel.com/changelog/private-storage-for-vercel-blob-now-available-in-public-beta

## What's New

Vercel Blob, the managed file storage service, now offers **private storage** capabilities in public beta. This addition allows developers to store sensitive files—such as contracts, invoices, and internal reports—with authentication-gated access, preventing accidental exposure through public URLs.

## Key Features

**Dual Storage Model:**
- **Public storage**: Allows public reads for media assets like images and videos
- **Private storage**: Requires authentication for all read and write operations

**Creating Private Stores:**
Create a new private store using the Vercel dashboard or CLI:
```bash
vercel blob create-store [name] --access private
```

When created within a linked Vercel project, the CLI automatically configures the `BLOB_READ_WRITE_TOKEN` environment variable for your deployments.

## Developer Implementation

**SDK Updates:**
Update to `@vercel/blob@2.3` or later to access private storage functionality.

**Uploading Files:**
Use the `put()` or `upload()` method with `access: 'private'`:
```typescript
const blob = await put(filename, request.body, {
  access: 'private',
});
```

**Retrieving Files:**
Stream private files using the `get()` method:
```typescript
const { stream, blob } = await get(filename, {
  access: 'private',
});
```

## Availability & Pricing

Private storage is available on all Vercel plans and uses standard Vercel Blob pricing. Full documentation is available in the [Vercel docs](https://vercel.com/docs/storage/vercel-blob/private-storage).

---

### Access billing usage and cost data via API
- Published: 2026-02-19T13:00:00+00:00
- Products: Vercel
- Tags: api, feature, integration
- URL: https://standarddb.com/vercel/vercel-adds-billing-api-with-focus-compliant-cost-data-access--nffmwzgm
- Original: https://vercel.com/changelog/access-billing-usage-cost-data-api

## New Billing API Endpoint

Vercel has launched a new `/billing/charges` API endpoint that enables programmatic access to billing usage and cost data. This addition addresses a key need for teams managing infrastructure costs across multiple services.

## Key Features

- **FOCUS v1.3 Compliance**: Data is returned in the FOCUS (FinOps Open Cost and Usage Specification) v1.3 format, an open standard that eliminates the need for custom transformation logic when integrating with FinOps platforms.
- **Flexible Date Ranges**: Supports 1-day granularity with a maximum date range of one year, allowing both recent cost analysis and historical retrospectives.
- **Efficient Data Streaming**: Responses are streamed as newline-delimited JSON (JSONL) to handle large datasets efficiently without memory constraints.

## Access Methods

Teams can query billing data through multiple interfaces:

- **SDK**: Use the Vercel TypeScript SDK with `vercel.billing.listBillingCharges()` for type-safe integration
- **REST API**: Direct HTTP requests to the `/v1/billing/charges` endpoint with bearer token authentication
- **CLI**: The `vercel usage` command provides quick access to billing data for the current period or custom date ranges

## Ecosystem Integration

Vantage has released a native integration that automatically syncs Vercel cost data to Vantage accounts, simplifying multi-tool cost observability without manual data exports.

Refer to the [API documentation](https://docs.vercel.com/docs/rest-api/reference/endpoints/billing/list-focus-billing-charges) and [CLI reference](https://vercel.com/docs/cli/usage) for implementation details.

---

### v0.16.3-rc2: install: prevent partial download script execution (#14311)
- Published: 2026-02-19T02:32:45+00:00
- Products: 
- Tags: release, bugfix, security
- URL: https://standarddb.com/ollama/ollama-v0-16-3-rc2-prevents-partial-script-execution-during-installation--yskkkgsp
- Original: https://github.com/ollama/ollama/releases/tag/v0.16.3-rc2

## Installation Script Safety Improvement

Ollama v0.16.3-rc2 addresses a potential security issue in the installation process by wrapping the download script in a main function. This change prevents scenarios where a partial or truncated download could execute incomplete code, which could lead to unpredictable behavior or security issues.

### What Changed

- Installation scripts are now wrapped in a `main()` function
- Truncated partial downloads will no longer execute incomplete code fragments
- Ensures that interrupted or incomplete downloads fail safely

### Why This Matters

When download scripts are executed directly, a network interruption or connection dropout mid-transfer could result in partial script execution. By wrapping the entire script in a main function, only complete, well-formed scripts will execute. Incomplete downloads simply won't invoke the main function and will terminate safely.

### Action Items

Users should update to v0.16.3-rc2 or later to benefit from this installation safety improvement. No action is required if your installation completes successfully, but this ensures better protection against installation-time failures.

---

### v0.5.94
- Published: 2026-02-19T01:13:24+00:00
- Products: Sim Studio
- Tags: release, feature, bugfix, integration
- URL: https://standarddb.com/sim/sim-studio-v0-5-94-adds-vercel-block-integration-and-next-js-16-upgrade--rzhctzyc
- Original: https://github.com/simstudioai/sim/releases/tag/v0.5.94

## New Features

Sim Studio v0.5.94 introduces **Vercel block and tools integration**, enabling developers to incorporate Vercel functionality directly into their workflows.

## Bug Fixes & Improvements

This release addresses several stability and usability issues:

- **Sidebar ordering**: Unified workflow and folder insertion ordering for consistent navigation
- **Shortlink redirects**: Fixed an issue where the `isHosted` guard was preventing redirects from functioning properly on ECS at build time

## Infrastructure Updates

The underlying technology stack has been strengthened with:

- **Next.js upgrade**: Updated from version 16.1.0-canary.21 to 16.1.6 for improved stability and performance

## Action Items

Users and developers should update to v0.5.94 to access the new Vercel integration capabilities and benefit from the latest Next.js stability improvements.

---

### AI Gateway, Workers AI - AI dashboard experience improvements
- Published: 2026-02-19T00:00:00+00:00
- Products: Cloudflare Workers, Cloudflare
- Tags: feature, platform, integration
- URL: https://standarddb.com/cloudflare/cloudflare-improves-ai-gateway-and-workers-ai-dashboard-with-better-navigation--yxzjhpwj
- Original: https://developers.cloudflare.com/changelog/post/2026-02-19-ai-dashboard-experience-improvements

## Dashboard Navigation

AI features now have their own dedicated top-level section in the Cloudflare dashboard sidebar, making it significantly easier to discover and access AI tools without navigating through multiple menu layers.

## Onboarding and Getting Started

The first-time setup experience for AI Gateway has been simplified with immediate visibility of your gateway's **OpenAI-compatible endpoint** and step-by-step guidance for configuration. The Playground now includes helpful prompts, and usage pages display clear next steps for users who haven't yet made requests.

Code examples have been consolidated into a single view with dropdown selectors for **API type, provider, SDK, and authentication method**, allowing developers to customize their exact code snippet from one place rather than navigating separate sections.

## Dynamic Routing Enhancements

- Route builder performance and responsiveness have been improved
- Single-click clipboard copy for route names
- Code examples now use the Universal Endpoint format for easier application integration

## Observability and Accessibility

**Analytics**: Small monetary values now display correctly in cost analytics charts, enabling accurate spending tracking at any scale.

**Accessibility**: Keyboard navigation improvements throughout AI Gateway (particularly for provider exploration) and enhanced sorting/filtering on the Workers AI models page make the interface more accessible to all users.

---

### A New Era of Collaboration for Legal and Professional Services
- Published: 2026-02-19T00:00:00+00:00
- Products: 
- Tags: feature, platform, api
- URL: https://standarddb.com/harvey/harvey-launches-shared-spaces-enabling-secure-cross-organizational--gycztzpv
- Original: https://www.harvey.ai/blog/a-new-era-of-collaboration-for-legal-and-professional-services

## Shared Spaces: Collaboration Beyond Organizational Boundaries

Harvey has launched **Shared Spaces**, addressing a critical gap in AI-powered legal collaboration. While earlier tools enabled document management and communication, this new feature brings AI-assisted legal work into shared, secure workspaces across organizational boundaries—allowing law firms, clients, and advisors to collaborate without compromising confidentiality.

## What Changed

The new capability extends Harvey's existing core features—shared bulk document analysis in Vault, streamlined review with Playbooks, and custom workflow automation—to external stakeholders. Teams can now invite clients, co-counsel, and advisors into intelligent workspaces where they collectively benefit from Harvey's AI-powered research, drafting, and document analysis tools.

## For Legal Teams

**Benefits include:**
- Secure cross-organizational collaboration on sensitive matters
- Unified access to AI-powered document analysis, drafting, and research across team boundaries
- New service model opportunities for law firms and professional service networks
- Maintained security and confidentiality standards for sensitive legal work

## Development Process

Shared Spaces was developed in close partnership with forward-thinking design partners—organizations willing to help shape how multiplayer, secure collaboration works in practice for AI-powered legal work. This collaborative approach ensures the feature meets real-world needs of law firms and professional service networks.

This update represents a significant expansion of Harvey's platform scope, moving from individual firm productivity to cross-organizational workflow enablement.

---

### Release: Slack CLI v3.13.0
- Published: 2026-02-19T00:00:00+00:00
- Products: 
- Tags: release, feature, sdk
- URL: https://standarddb.com/slack/slack-cli-v3-13-0-improves-developer-experience-with-agentic-app-templates-and--spnmrvlc
- Original: https://docs.slack.dev/changelog/2026/02/19/slack-cli

## Agentic App Template Improvements

When running `slack create agent`, the CLI now prominently displays available agentic app templates for easier selection. If you want to name your app "agent" without creating an agentic app, you can use the `--name` flag to explicitly set the app name instead.

## Enhanced Template Browsing

The `slack create` command with the `--list` flag now displays available templates without requiring an immediate selection, allowing developers to explore options before committing.

## Deployment and Authentication Enhancements

- The `slack deploy` command now references the latest online documentation and provides helpful suggestions to run a development app using `slack run` if a `deploy` hook script doesn't exist.
- The CLI now automatically prompts for login when running credential-required commands and no prior authentications are detected, eliminating confusing error exits.

These improvements streamline the developer workflow from app creation through deployment and reduce friction when managing authentication.

---

### Osprey: Open Sourcing our Rule Engine
- Published: 2026-02-19T00:00:00+00:00
- Products: 
- Tags: open-source, release, feature, platform, api
- URL: https://standarddb.com/discord/discord-open-sources-osprey-rule-engine-for-real-time-safety-detection--tsnklrjy
- Original: https://discord.com/blog/osprey-open-sourcing-our-rule-engine

## Overview

Discord is open-sourcing **Osprey**, its rule engine for trust and safety operations, in collaboration with ROOST and the internet.dev team. Osprey enables platforms to investigate real-time user activities and deploy dynamic rules to address emerging threats, significantly reducing the engineering overhead typically required to build safety infrastructure from scratch.

## How Osprey Works

Osprey processes **Actions** (events in JSON format) through a series of configurable **Rules** written in SML (Some Made-up Language), a Python-inspired rule syntax designed to be accessible to non-technical safety teams. The system leverages several core components:

- **Rules:** Written in SML, rules can reference other rules and external data sources. Example: detecting known spammers based on email patterns and applying automatic labels.
- **UDFs (User Defined Functions):** Python functions that extend Osprey's capabilities, enabling integrations with external services like spam detection models.
- **Features & Entities:** Named variables extracted from actions and stored for later analysis. Entities are persistent units (users, servers, emails) that can have effects applied to them.
- **Effects:** Actions triggered when rules evaluate to true, such as labeling an entity as a spammer or blocking activity.

## Key Capabilities

Osprey is designed to handle the scale and speed requirements of large platforms:

- **Real-time processing:** Handles thousands of events per second via synchronous GRPC or asynchronous message queues
- **Rapid rule deployment:** New rules take effect in minutes, enabling quick response to emerging threats
- **Full transparency:** All rule executions and verdicts are logged and indexed in Apache Druid for investigation and debugging
- **Extensibility:** Support for custom UDFs and validation logic allows teams to adapt to new attack patterns
- **Continuous feedback loop:** Detection insights feed back into rule improvement

## Getting Started

The engine is available on GitHub at [github.com/roostorg/osprey](https://github.com/roostorg/osprey). Teams can use Osprey to build stronger safety measures by defining custom rules, integrating external data sources, and analyzing execution outputs through the investigation UI.

---

### February 19, 2026
- Published: 2026-02-19T00:00:00+00:00
- Products: Claude
- Tags: api, feature, deprecation
- URL: https://standarddb.com/anthropic/claude-launches-automatic-prompt-caching-and-retires-sonnet-3-7-haiku-3-5-models--dctqclpw
- Original: https://docs.anthropic.com/en/release-notes/overview

## Automatic Prompt Caching

Claude's Messages API now supports **automatic caching** without requiring manual breakpoint management. Developers can add a single `cache_control` field to their request body, and the system automatically caches the last cacheable block while moving the cache point forward as conversations grow. This feature works alongside existing block-level cache control for fine-grained optimization and is available on the Claude API and Azure AI Foundry (preview).

## Model Retirements and Upgrades

Anthropic has retired two models effective immediately:
- **Claude Sonnet 3.7** (`claude-3-7-sonnet-20250219`) — upgrade to Claude Sonnet 4.6
- **Claude Haiku 3.5** (`claude-3-5-haiku-20241022`) — upgrade to Claude Haiku 4.5

All requests to these models will now return errors. Additionally, **Claude Haiku 3** (`claude-3-haiku-20240307`) has been marked for deprecation with retirement scheduled for April 19, 2026.

Researchers requiring ongoing access to retired models can request access through the [External Researcher Access Program](https://support.claude.com/en/articles/9125743-what-is-the-external-researcher-access-program).

## Action Items for Developers

Update your API requests to use the current model versions:
- Replace `claude-3-7-sonnet-20250219` with `claude-sonnet-4.6`
- Replace `claude-3-5-haiku-20241022` with `claude-haiku-4.5`
- Plan migration of `claude-3-haiku-20240307` before April 19, 2026

---

### Security Center - Cloudforce One Threat events graphs are now visible in the dashboard
- Published: 2026-02-19T00:00:00+00:00
- Products: Cloudflare
- Tags: feature, platform
- URL: https://standarddb.com/cloudflare/cloudflare-adds-threat-event-visualizations-to-security-center-dashboard--kcxtcxly
- Original: https://developers.cloudflare.com/changelog/2026-02-19-threat-events-graphs

## New Threat Event Visualizations

Cloudflare has enhanced the Threat Events dashboard with interactive data visualizations designed to help security teams understand attack patterns and threat landscapes more effectively.

### Key Features

**Sankey Diagrams** trace the flow of attacks from origin country to target country, making it easy to identify which regions face the heaviest threat activity and where attack infrastructure originates.

**Dataset Distribution over Time** enables instant pivoting to determine whether specific campaigns target your sector or represent broad-spectrum commodity attacks affecting multiple organizations.

**Enhanced Filtering** allows users to filter and drill down into specific attack vectors directly from the charts, streamlining threat investigation workflows.

### Availability

These visualizations are now available to Cloudforce One subscribers in the Cloudflare dashboard under **Application Security > Threat Intelligence > Threat Events**. No additional configuration is required—the new views are active immediately.

### Use Cases

Security teams can leverage these tools to:
- Correlate attack origins and targets to understand geopolitical threat patterns
- Distinguish between targeted and indiscriminate attacks
- Quickly isolate and investigate specific threat campaigns


---

### Deploying DeepSeek on GB300 NVL72: Big Wins in Long-Context Inference
- Published: 2026-02-19T00:00:00+00:00
- Products: 
- Tags: release, feature, performance, integration, api
- URL: https://standarddb.com/lmsys/nvidia-and-sglang-optimize-deepseek-for-gb300-nvl72-achieving-226-tokens-per--zsvzgrjz
- Original: https://lmsys.org/blog/2026-02-19-gb300-longctx

## Long-Context Inference Optimization for GB300

NVIDIA and the SGLang team have released production-ready optimizations for deploying DeepSeek R1 on the GB300 NVL72 GPU cluster. The work focuses on long-context inference scenarios (128K input / 8K output tokens) where throughput and latency are critical performance metrics.

## Key Performance Gains

- **Peak throughput**: 226.2 tokens per second per GPU on GB300, representing a **1.53x advantage** over GB200 configurations
- **Under matched latency**: GB300 delivers **1.38x–1.58x** tokens per second compared to GB200 across representative workloads
- **Per-user throughput**: Multi-token prediction enables an **1.87x increase** in tokens per second per user
- **TTFT improvement**: 8.6 seconds for 128K token prefill, **1.07x–1.23x faster** than GB200, driven by optimized attention kernels

## Technical Approaches

The optimization employs several complementary techniques:

**Prefill optimizations**: Chunked pipeline parallelism distributes long-context prompt processing across pipeline stages with dynamic chunking. The team enabled FP8 attention and FP8 KV-cache support to reduce memory traffic and double KV-cache capacity within fixed memory footprint. GB300's upgraded Special Function Unit (SFU) provides 2x throughput for softmax operations, yielding a 1.35x speedup in the FMHA kernel.

**Decode optimization**: Expert parallelism (Wide-EP) distributes MoE weights and KV cache across up to 32 GPUs, reducing per-GPU memory pressure. GB300's 288 GB of HBM3e (versus GB200's 192 GB) directly enables 1.6x higher effective decode batch size—40 vs 24 requests per GPU at DEP8.

**Orchestration**: The deployment uses NVIDIA Dynamo, a control plane for cluster-scale disaggregated inference that handles KV-cache-aware routing and worker coordination with near-zero scheduling overhead.

## Reproduction and Production Deployment

Complete reproduction instructions are available in the SGLang GitHub repository (issue #18703). For production deployments, the Dynamo Kubernetes stack offers GB200/GB300 support with inference-aware autoscaling and cluster topology-aware scheduling for disaggregated inference workloads.

---

### AI Gateway, Workers AI - AI dashboard experience improvements
- Published: 2026-02-19T00:00:00+00:00
- Products: Cloudflare Workers, Cloudflare
- Tags: feature, platform, integration
- URL: https://standarddb.com/cloudflare/cloudflare-improves-ai-gateway-and-workers-ai-dashboard-experience-with--cvxyfwfg
- Original: https://developers.cloudflare.com/changelog/2026-02-19-ai-dashboard-experience-improvements

## Dashboard Navigation and Accessibility

Cloudflare has given its AI products dedicated prominence in the dashboard with a new top-level AI section in the sidebar navigation. This makes AI Gateway and Workers AI features more discoverable without requiring developers to navigate through multiple menu levels. The update also includes improved keyboard navigation and sorting/filtering components, particularly on the Workers AI models page, making the platform more accessible to all users.

## Streamlined Onboarding

Getting started with AI Gateway is now faster with an improved onboarding flow. When creating your first gateway, developers see:
- Their gateway's OpenAI-compatible endpoint immediately
- Step-by-step configuration guidance
- Helpful prompts in the Playground
- Clear next steps if no requests have been made yet

Code examples have been consolidated into a single view with dropdown selectors for API type, provider, SDK, and authentication method, letting developers customize exactly what they need from one place.

## Route Builder and Analytics Improvements

The dynamic routing feature receives several enhancements:
- **Performance**: The route builder is now more responsive and performant
- **Usability**: One-click copying of route names to clipboard
- **Integration**: Code examples now use the Universal Endpoint format for easier application integration

Cost analytics have also been refined to accurately display small monetary values in charts, enabling precise spending tracking at any scale.

For more details, check out the [AI Gateway documentation](https://developers.cloudflare.com/ai-gateway/).

---

### Security Center - Cloudforce One Threat events graphs are now visible in the dashboard
- Published: 2026-02-19T00:00:00+00:00
- Products: Cloudflare
- Tags: feature, security, platform
- URL: https://standarddb.com/cloudflare/cloudflare-adds-threat-visualization-to-cloudforce-one-dashboard-with-sankey--tpfptydr
- Original: https://developers.cloudflare.com/changelog/post/2026-02-19-threat-events-graphs

## New Threat Events Visualizations

Cloudflare has rolled out dynamic visualization tools to the Cloudforce One Threat Events dashboard, designed to help security teams better understand and analyze threat landscapes in real time.

### Key Features

**Sankey Diagrams** provide visual tracing of attack flows from country of origin to target location. This helps identify which geographic regions face the heaviest attack volume and where threat infrastructure is concentrated.

**Dataset Distribution Charts** allow teams to pivot their view over time, making it easy to determine whether a specific campaign targets your sector or represents broad-spectrum commodity attacks affecting multiple organizations.

**Enhanced Filtering Capabilities** let you leverage these visual tools to filter and drill down into specific attack vectors directly from the charts, enabling faster threat investigation and response.

### How to Access

Cloudforce One subscribers can access these new visualizations in the Cloudflare dashboard under **Application Security > Threat Intelligence > Threat Events**. These tools are available immediately to eligible accounts.

---

### Changes to test merge commit generation for pull requests
- Published: 2026-02-19T00:00:00+00:00
- Products: GitHub
- Tags: feature, performance, platform
- URL: https://standarddb.com/github/github-optimizes-test-merge-commit-generation-for-pull-requests-reducing-delays--qytntxsq
- Original: https://github.blog/changelog/2026-02-19-changes-to-test-merge-commit-generation-for-pull-requests

## Test Merge Commit Generation Changes

GitHub has optimized the frequency of test merge commit generation for pull requests to reduce delays and improve system reliability. This backend change focuses on when these commits are created, not how mergeability is determined.

### When Test Merge Commits Are Generated

Test merge commits are now only generated under these conditions:

- **Code changes**: When new commits are pushed to the pull request branch
- **Merge base changes**: When the merge base between the PR and base branches changes
- **Time-based refresh**: When the current test merge commit is older than 12 hours

Previously, test merge commits were regenerated every time a pull request page was viewed, causing unnecessary system load.

### What Remains Unchanged

This optimization does not affect:

- **Mergeability checks**: PR mergeability status determination remains unaffected
- **Merge conflict detection**: Conflicts are still properly reported
- **Rule enforcement**: Branch protection and merge rules continue to work as expected

### Impact for Developers

This change improves the overall responsiveness of GitHub's pull request system by reducing computational overhead. Developers will experience faster page load times and more reliable mergeability status updates, with no impact on the accuracy or completeness of merge conflict detection and rule enforcement.

---

### GitHub Projects: Import items based on a query and hierarchy view improvements
- Published: 2026-02-19T00:00:00+00:00
- Products: GitHub
- Tags: feature, platform
- URL: https://standarddb.com/github/github-projects-adds-search-based-imports-and-hierarchy-view-improvements--gyxygdrs
- Original: https://github.blog/changelog/2026-02-19-github-projects-import-items-based-on-a-query-and-hierarchy-view-improvements

## Import Project Items with Search Queries

GitHub Projects now enables users to populate new projects using advanced search queries, expanding beyond direct repository imports. The search functionality supports all filters available on the Issues page, including `AND` / `OR` keywords and nested queries.

This capability allows teams to create projects with precisely curated item sets without manual selection, streamlining project initialization for complex filtering scenarios.

## Hierarchy View Enhancements

Following the public preview launch in January, GitHub Projects has shipped several improvements to its hierarchy view based on user feedback:

- **Inline sub-issue management**: Create and add sub-issues directly from within a project
- **Drag-and-drop reordering**: Reorganize and reparent sub-issues using intuitive drag-and-drop interactions
- **Synchronized ordering**: Sub-issue ordering now stays consistent between the issues view and projects view

## Upcoming Features

GitHub is committed to further enhancing hierarchy view with planned improvements including:

- Preventing duplicate issue display when items already appear in the hierarchy
- Preserving expand/collapse states across sessions
- Adding filtering capabilities for sub-issues (e.g., displaying only open sub-issues)

Users can participate in discussions about these features on the [GitHub Community forum](https://github.com/orgs/community/discussions/186435).


---

### Access all pull request comments without leaving the new Files changed page
- Published: 2026-02-19T00:00:00+00:00
- Products: GitHub
- Tags: feature, platform
- URL: https://standarddb.com/github/github-brings-pull-request-comments-to-files-changed-page-eliminating-tab--bdzkxsgl
- Original: https://github.blog/changelog/2026-02-19-access-all-pull-request-comments-without-leaving-the-new-files-changed-page

## Overview

Code review workflows just got more efficient. GitHub has integrated general pull request comments directly into the Files changed page, eliminating the need to switch between the "Files changed" and "Conversation" tabs to understand full context.

## Key Changes

**Comments Panel Integration**
- A new **Comments panel** now displays all pull request-level comments alongside your code review
- Review code and follow high-level conversations in a single view
- Add new comments directly from the Comments panel without navigating away

**Enhanced Reply Capabilities**
- **Quote reply** functionality allows you to respond to existing comments with context preserved
- Discussions remain clear and organized with proper threading

**Streamlined Review Process**
- When requesting changes, you no longer need to provide a summary comment if you've already left inline comments
- Submit reviews more quickly without redundant explanations

**Improved Filtering**
- Resolved and unresolved comment filters now display states correctly
- Filter updates work as expected throughout the review lifecycle

## What Developers Need to Do

No action required—these improvements are automatically available in the new Files changed page preview. Start using the integrated Comments panel during your next code review to keep discussions and code changes in one focused workspace.

---

### Digital Experience Monitoring - DEX Supports EU Customer Metadata Boundary
- Published: 2026-02-19T00:00:00+00:00
- Products: Cloudflare
- Tags: feature, api, platform, security
- URL: https://standarddb.com/cloudflare/cloudflare-dex-now-complies-with-eu-customer-metadata-boundary-settings--gsqnysvh
- Original: https://developers.cloudflare.com/changelog/post/2026-02-19-dex-supports-cmb-eu

## EU Data Residency Support for DEX

Cloudflare has updated its Digital Experience Monitoring (DEX) service to fully comply with the Customer Metadata Boundary (CMB) setting for the European Union. This ensures that DEX logs—which track WARP device connectivity and performance metrics—will not be stored outside the EU when CMB is configured.

## What This Means for EU Customers

When a Cloudflare One customer enables CMB for the EU region, DEX data will no longer appear in the Cloudflare One dashboard. Instead, a notification message will inform users: "DEX data is unavailable because Customer Metadata Boundary configuration is on. Use Cloudflare LogPush to export DEX datasets."

## Data Access via LogPush

EU customers who need to maintain DEX visibility can leverage **LogPush** to export DEX datasets directly. This allows organizations to:
- Build custom analytics and dashboards with their DEX data
- Maintain full control over data processing and storage
- Comply with EU data residency requirements while retaining operational visibility

This update is particularly relevant for enterprises operating in the EU that require strict data localization compliance while using WARP and DEX for network performance monitoring.

---

### February 19, 2026
- Published: 2026-02-19T00:00:00+00:00
- Products: Claude
- Tags: release, feature, api, model, deprecation
- URL: https://standarddb.com/anthropic/anthropic-launches-automatic-prompt-caching-and-claude-sonnet-4-6-retires--hmbcxlcw
- Original: https://platform.claude.com/docs/en/release-notes/overview

## New Features

**Automatic Prompt Caching** is now available on the Claude API and Azure AI Foundry (preview). Developers can add a single `cache_control` field to their request body and the system automatically caches the last cacheable block, advancing the cache point as conversations grow. This eliminates manual breakpoint management while working alongside existing block-level cache control for fine-grained optimization.

**Claude Sonnet 4.6** is Anthropic's latest balanced model, combining speed and intelligence for everyday tasks. It delivers improved agentic search performance while consuming fewer tokens. The model supports extended thinking and a 1M token context window (beta).

## Model Retirements & Deprecations

- **Retired immediately**: Claude Sonnet 3.7 (`claude-3-7-sonnet-20250219`) and Claude Haiku 3.5 (`claude-3-5-haiku-20241022`) — all requests now return errors. Users should upgrade to Claude Sonnet 4.6 and Claude Haiku 4.5 respectively.
- **Deprecated with sunset date**: Claude Haiku 3 (`claude-3-haiku-20240307`) will be retired on April 19, 2026. Migration to Claude Haiku 4.5 is recommended.
- Researchers can request ongoing access through the [External Researcher Access Program](https://support.claude.com/en/articles/9125743-what-is-the-external-researcher-access-program).

## Developer Actions Required

Developers using the retired models (Sonnet 3.7, Haiku 3.5) must immediately update their API calls to use the recommended replacement models. For Haiku 3, migration should be planned before April 19, 2026.

---

### Workflow dispatch API now returns run IDs
- Published: 2026-02-19T00:00:00+00:00
- Products: GitHub Actions
- Tags: api, feature, release
- URL: https://standarddb.com/github/github-actions-workflow-dispatch-api-now-returns-run-ids-and-urls--jgdjyxtc
- Original: https://github.blog/changelog/2026-02-19-workflow-dispatch-api-now-returns-run-ids

## What Changed

The GitHub Actions [workflow dispatch API endpoint](https://docs.github.com/rest/actions/workflows#create-a-workflow-dispatch-event) now supports returning workflow run metadata. Previously, this endpoint returned only a `204 No Content` status code with no response body, making it difficult to programmatically identify which workflow runs originated from API-triggered events.

## New Capabilities

- **Optional `return_run_details` parameter**: Pass this boolean parameter in your API request to receive a `200 OK` response containing:
  - Workflow ID
  - API URL for the workflow run
  - Workflow run URL
- **Backward compatible**: Omitting the parameter preserves the current `204 No Content` behavior
- **CLI support**: Available in GitHub CLI v2.87.0+, with `gh workflow run` now returning the created run URL and providing quick access via `gh run view` commands

## Developer Impact

With this update, developers can eliminate manual tracking solutions and extensive polling to map API requests to workflow runs. The new parameter defaults to `true` in the latest GitHub CLI version, streamlining the developer experience when triggering workflows programmatically.

Refer to the [GitHub Actions documentation](https://docs.github.com/actions/using-workflows/events-that-trigger-workflows#workflow_dispatch) for detailed implementation guidance.

---

### Pull request throughput and time to merge available in Copilot usage metrics API
- Published: 2026-02-19T00:00:00+00:00
- Products: GitHub Copilot, GitHub Dependabot
- Tags: api, feature, platform
- URL: https://standarddb.com/github/github-copilot-usage-metrics-api-now-tracks-pull-request-throughput-and-merge--dxwtlyjf
- Original: https://github.blog/changelog/2026-02-19-pull-request-throughput-and-time-to-merge-available-in-copilot-usage-metrics-api

## New Metrics Available

GitHub's Copilot usage metrics API now includes enterprise-level metrics focused on pull request outcomes:

- **Pull request review suggestions and acceptance rates** — Track how many review suggestions generated by Copilot are acted upon
- **Copilot-generated PRs that get merged** — Measure the success rate of pull requests created with Copilot's coding agent
- **Pull request cycle time** — Monitor the time from PR creation to merge, identifying velocity improvements

## Who Can Access

This feature is available to enterprises that have enabled Copilot usage metrics. Access requires either enterprise admin privileges or the new `View Enterprise Copilot Metrics` role, ensuring appropriate governance over metrics visibility.

## Key Benefits

Organizations can now:

- **Measure developer productivity** — Quantify how Copilot influences PR throughput and review efficiency across teams
- **Track time-to-merge improvements** — Identify whether Copilot reduces the cycle time from idea to merged code
- **Validate adoption impact** — Use concrete metrics to justify Copilot investments and guide rollout decisions

The API integrates with GitHub's existing Copilot usage metrics infrastructure, allowing enterprises to build custom dashboards and reports on top of these endpoints.

---

### Use Copilot coding agent with Windows projects
- Published: 2026-02-18T22:04:40+00:00
- Products: GitHub Copilot, GitHub Actions
- Tags: feature, platform
- URL: https://standarddb.com/github/github-copilot-coding-agent-now-supports-windows-development-environments--mwxjgrfg
- Original: https://github.blog/changelog/2026-02-18-use-copilot-coding-agent-with-windows-projects

## Windows Environment Support

GitHub Copilot's autonomous coding agent—which delegates tasks to run asynchronously in the background—previously defaulted to Linux environments. Developers targeting Windows projects can now configure the agent to use a Windows development environment instead, maintaining the same GitHub Actions-powered execution model.

## Configuration and Setup

To enable Windows support, create a `copilot-setup-steps.yml` file in your repository and specify a custom `runs-on` setting to direct the agent to use Windows runners. This allows the agent to execute in an environment that matches your project's actual deployment target.

## Build Verification and Quality Gates

With a properly configured Windows development environment, Copilot can ensure that:

- Your project builds successfully
- Automated tests pass
- Linter checks complete without errors

This reduces the likelihood of failed or problematic pull requests, letting you merge with confidence on the first attempt.

## Important Limitations and Recommendations

The integrated firewall feature that Copilot includes is not compatible with Windows environments. GitHub recommends using Windows configurations only with:

- Self-hosted runners
- Larger runners with Azure private networking where you can implement custom network controls

For implementation details, see the GitHub Docs section on ["Customizing the development environment for Copilot coding agent"](https://docs.github.com/copilot/how-tos/use-copilot-agents/coding-agent/customize-the-agent-environment#switching-copilot-to-a-windows-development-environment).

---

### Copilot coding agent supports code referencing
- Published: 2026-02-18T22:02:22+00:00
- Products: GitHub Copilot
- Tags: feature, api
- URL: https://standarddb.com/github/github-copilot-coding-agent-now-supports-code-referencing-with-public--xwnlvyvk
- Original: https://github.blog/changelog/2026-02-18-copilot-coding-agent-supports-code-referencing

## Code Referencing Now Available in Copilot Coding Agent

GitHub has extended its **code referencing** feature to the Copilot coding agent, enabling the autonomous background agent to track and surface code matches from public repositories. When the agent generates code that matches existing public code on GitHub, the matching code is highlighted directly in the agent session logs with a link to the original source.

### What This Means for Developers

This addition provides critical transparency into code origins:

- **Visibility into matches**: Generated code that matches public repositories is flagged in session logs
- **License awareness**: Links to original code help developers understand applicable licenses
- **Provenance tracking**: Clear attribution for where suggestions originate

### Important Policy Behavior

The feature has one key limitation: **the "Block" mode from the "Suggestions matching public code" policy is not enforced** in the coding agent. When a policy is set to Block, matching code suggestions will not be prevented; instead, they will be highlighted in session logs for transparency.

Developers relying on strict blocking behavior should be aware of this exception when using the autonomous agent.

### Next Steps

For detailed information on configuring and using this feature, see the GitHub documentation on [finding and viewing code references for the Copilot coding agent](https://docs.github.com/copilot/how-tos/get-code-suggestions/find-matching-code?tool=webui#view-code-references-for-copilot-coding-agent).

---

### v0.5.93
- Published: 2026-02-18T20:32:03+00:00
- Products: Sim Studio
- Tags: release, feature, security, bugfix
- URL: https://standarddb.com/sim/sim-releases-v0-5-93-with-audit-logging-and-enterprise-access-controls--ymvqmkfr
- Original: https://github.com/simstudioai/sim/releases/tag/v0.5.93

## Audit Logging & Compliance

Sim v0.5.93 ships a **persistent audit log system** with comprehensive route instrumentation to track user actions across the platform. The audit log now captures events for templates, billing, credentials, environment variables, deployments, and password changes—essential for enterprises requiring compliance tracking and security visibility.

## Enterprise Access Controls

The release adds several security-focused features for enterprise deployments:

- **Usage limit visibility**: Non-admin enterprise members can no longer see usage limits and seats information, restricting sensitive billing data to administrators
- **Integration allowlisting**: A new `ALLOWED_INTEGRATIONS` environment variable enables self-hosted operators to restrict which integration blocks are available
- **MCP domain allowlisting**: The `ALLOWED_MCP_DOMAINS` environment variable allows administrators to control which domains MCP can connect to

## Canvas & Locking Improvements

Locked blocks in the canvas now support outbound connections, providing more flexibility for workflow design. A critical fix prevents socket crashes when locking agent blocks, improving stability during collaborative editing.

## Additional Fixes

The patch also addresses a Copilot shortcut conflict, updates integration normalization checks to handle lowercase values, and fixes redirect handling for Beluga tracking in shortlinks.

---

### Claude Opus 4.6 is now available in Visual Studio, JetBrains IDEs, Xcode, and Eclipse
- Published: 2026-02-18T20:05:54+00:00
- Products: GitHub Copilot
- Tags: release, feature, model, integration
- URL: https://standarddb.com/github/claude-opus-4-6-now-available-in-github-copilot-across-visual-studio-jetbrains--zmjvwkkp
- Original: https://github.blog/changelog/2026-02-18-claude-opus-4-6-is-now-available-in-visual-studio-jetbrains-ides-xcode-and-eclipse

## Broader IDE Support for Claude Opus 4.6

Claude Opus 4.6 is now generally available within GitHub Copilot, extending access to developers using a wide range of development environments. The model can be accessed through GitHub Copilot Chat on github.com, GitHub Mobile, Visual Studio Code, Visual Studio, JetBrains IDEs, Xcode, and Eclipse.

## How to Access

Users can select Claude Opus 4.6 directly from the chat model picker in **agent, ask, and edit modes**. The availability applies to:
- **Copilot Enterprise** subscribers
- **Copilot Business** subscribers  
- **Copilot Pro** subscribers
- **Copilot Pro+** subscribers

For Copilot Business and Enterprise organizations, administrators must opt in by enabling the Claude Opus 4.6 policy in Copilot settings before users will see the model in their IDE.

## What This Means for Developers

This release expands model choice within GitHub Copilot's ecosystem, giving developers using Anthropic's Claude model access across their preferred IDEs without switching tools. The multi-IDE rollout makes Claude Opus 4.6 immediately practical for teams with heterogeneous development environments, including iOS developers using Xcode, Java teams relying on JetBrains IDEs, and traditional .NET developers working in Visual Studio.

For more details on available models and features, see GitHub's [Copilot models documentation](https://docs.github.com/copilot/reference/ai-models/supported-models).

---

### A new way to express yourself: Gemini can now create music
- Published: 2026-02-18T16:01:38+00:00
- Products: Gemini
- Tags: feature, model, release
- URL: https://standarddb.com/google/gemini-adds-music-generation-with-lyria-3-model-creates-30-second-tracks-from--dzxrhshc
- Original: https://deepmind.google/blog/a-new-way-to-express-yourself-gemini-can-now-create-music

## Music Generation Now Available in Gemini

Google DeepMind has launched **Lyria 3**, its latest generative music model, in beta within the Gemini app. The feature empowers users to create custom 30-second music tracks either by typing a description or uploading a photo. Whether you want "a comical R&B slow jam about a sock finding their match" or a personalized song based on family memories, Gemini translates your input into high-quality, catchy audio in seconds.

## Key Improvements Over Previous Models

Lyria 3 represents a significant upgrade to Google's music generation capabilities:

- **Automatic Lyric Generation**: No need to write or provide lyrics—the model generates them based on your text prompt or visual inspiration
- **Enhanced Creative Control**: Users can specify style, vocal characteristics, and tempo preferences for more personalized results
- **Improved Audio Quality**: Creates more realistic and musically complex tracks compared to earlier versions

## How to Use the Feature

The feature supports two primary workflows:

1. **Text-to-Track**: Describe a genre, mood, memory, or specific request to generate unique tracks with lyrics or instrumental audio
2. **Photo/Video-to-Track**: Upload visual content and Gemini composes a track with lyrics that match the mood and context

Each generated track includes custom cover art created by Nano Banana, making it easy to share via download or shareable links.

## Additional Details and Availability

All music generated through Gemini is embedded with **SynthID**, Google's imperceptible watermark technology that identifies AI-generated audio. This addresses concerns around AI-generated content attribution and verification.

The feature is available in beta in the Gemini app for users 18 and older across multiple languages, with higher generation limits for paid subscribers. Lyria 3 is also available to YouTube creators through YouTube's Dream Track feature, which is rolling out globally to enhance Shorts soundtracks.

---

### New in Agent Builder: all new agent chat, file uploads + tool registry
- Published: 2026-02-18T15:55:08+00:00
- Products: 
- Tags: feature, platform, api
- URL: https://standarddb.com/langchain/langsmith-agent-builder-adds-unified-chat-interface-file-uploads-and-tool--lzhgfgxr
- Original: https://blog.langchain.com/new-in-agent-builder-all-new-agent-chat-file-uploads-tool-registry

## New Chat Agent with Full Tool Access

LangSmith Agent Builder now includes an always-available Chat agent that can access every connected tool in your workspace—including Slack, Gmail, Linear, Pylon, and remote MCP servers. This eliminates the need to create specialized agents for one-off tasks. Users can ask natural language questions like "What are my open Linear tickets?" or "Summarize today's support requests," and the agent will plan, execute, and request approval before taking sensitive actions.

The Chat agent can orchestrate multiple tools in sequence. For example, you can ask it to summarize the last 10 emails, then automatically create calendar events for any meeting requests it finds. Multiple independent chat threads can run simultaneously, allowing users to manage multiple workflows and return to any conversation to approve actions or ask follow-up questions.

## Convert Conversations Into Reusable Agents

A major workflow improvement lets you turn any chat conversation directly into a recurring agent with a single click. Rather than requiring prompt engineering or conditional logic, the conversation itself becomes the agent's setup. This pattern works particularly well for research, content writing, summarization, and outreach workflows. Created agents can be triggered manually, scheduled on a cadence, or configured to respond to external events like Slack messages or emails.

## File Upload Capabilities

Users can now upload CSVs, images, and text files directly into chat for data analysis, image processing, and reference material. For example, uploading a CSV of sales data lets the agent analyze trends and share reports; uploading a screenshot enables conversion to formatted documents; importing style guides or existing prompts helps bootstrap new agent configurations.

## Centralized Tool Management

A new workspace tool registry provides a single location to view all connected tools, authenticate new integrations, and manage remote MCP servers. Re-authentication needs are surfaced immediately. Only workspace administrators can add tools, providing governance controls while making new tools instantly available to all agents in the workspace.

---

### Practice any skill, any time in the Practice tab
- Published: 2026-02-18T13:00:00+00:00
- Products: 
- Tags: feature, release, platform
- URL: https://standarddb.com/duolingo/duolingo-launches-free-practice-tab-for-on-demand-skill-targeting-across-all--kcklpvwg
- Original: https://blog.duolingo.com/guide-to-duolingo-practice-hub

## New Practice Tab Available to All Learners

Duolingo has launched a **Practice tab** that enables learners to break away from their regular lesson sequence and focus on skills that matter most to them. Available on iOS with Android coming soon, the feature is free for all users and complements the platform's existing curriculum.

## Practice Options

The Practice tab offers several distinct training modes:

- **Mistakes**: Review exercises where learners previously made errors, helping them avoid repeating the same mistakes
- **Words**: Study recommended vocabulary to strengthen language comprehension
- **Speak**: Practice pronunciation and speaking confidence with guided exercises
- **Listen**: Develop listening comprehension skills by identifying words and phrases from audio prompts
- **Focused Listening**: A featured practice mode combining audio recognition

The tab also integrates access to other Duolingo content including Stories, Radio, and Adventures for immersive practice. Max subscribers can additionally review Video Call and Roleplay sessions.

## Access and Rollout

To access the Practice tab, users can tap the barbell icon in the bottom menu on their home screen, or find it in the extended menu (three-dot menu) if not immediately visible. The feature rolled out initially on iOS for all language courses, with Android support expected soon.

## Alignment with Duolingo's Accessibility Mission

This launch follows Duolingo's earlier decision to make "Explain My Answer"—an in-lesson feedback feature—free to all learners. The Practice tab represents another step toward the company's stated mission of making quality education universally available.

---

### Redesigned search and filtering for runtime logs
- Published: 2026-02-18T13:00:00+00:00
- Products: Vercel
- Tags: feature, platform
- URL: https://standarddb.com/vercel/vercel-redesigns-runtime-logs-search-with-structured-filters-and-smart--jrpbsclp
- Original: https://vercel.com/changelog/redesigned-search-and-filtering-for-runtime-logs

## Improved Log Filtering Experience

Vercel has redesigned the Runtime Logs search bar in the project dashboard to streamline log exploration and filtering. The update focuses on making complex queries more manageable and intuitive for developers working with production logs.

### Key Features

**Structured Filters**: Queries like `level:error` or `status:500` are now parsed into visual pills that display at a glance. Each filter can be easily removed with a single click, making it simple to edit and refine searches without retyping entire queries.

**Smarter Suggestions**: The search bar now provides intelligent suggestions as you type, pulling filter values from your actual log data. Recent queries are automatically saved per-project and displayed at the top for quick access to commonly run searches.

**Better Input Handling**: Real-time validation flags typos and errors with helpful tooltips before you execute a search. The search bar also recognizes Vercel Request IDs and automatically converts them into filters for faster debugging.

These improvements are available immediately in your project dashboard and should help reduce friction when diagnosing issues with your deployed applications. See the [runtime logs documentation](https://vercel.com/docs/observability/runtime-logs) for more details.

---

### npm bulk trusted publishing config and script security now generally available
- Published: 2026-02-18T00:00:00+00:00
- Products: GitHub
- Tags: release, security, api, deprecation
- URL: https://standarddb.com/github/npm-cli-v11-10-0-adds-bulk-oidc-publishing-and-git-dependency-security-controls--wgwlkdyp
- Original: https://github.blog/changelog/2026-02-18-npm-bulk-trusted-publishing-config-and-script-security-now-generally-available

## Bulk OIDC Trusted Publishing Configuration

Maintainers can now configure OIDC trusted publishing across multiple packages simultaneously using the new [`npm trust`](https://docs.npmjs.com/cli/v11/commands/npm-trust) command. This eliminates the need to individually configure each package, streamlining the process of setting up automated, keyless authentication for package publishing.

## Git Dependency Security Controls

A critical new security feature addresses a supply chain vulnerability: git dependencies—whether direct or transitive—can include `.npmrc` files that override the git executable path, potentially enabling arbitrary code execution during `npm install` even when using `--ignore-scripts`. 

The new **`--allow-git` flag** provides explicit control over this behavior:

- `--allow-git=all` (current default): Allows all git dependencies
- `--allow-git=none`: Blocks git dependencies entirely
- Additional granular options available for finer control

**Recommended action**: Start using `--allow-git=none` in your workflows now. The npm team expects this to become the default in npm CLI v12, so adopting it early prevents future disruptions.

## Migration Path

While the current default is `all` for backward compatibility, developers should audit their dependency trees and explicitly set `--allow-git=none` unless git dependencies are genuinely required. The shift to a secure-by-default posture represents a meaningful change in how npm handles supply chain risk.

---

### Split Diffs are Here
- Published: 2026-02-18T00:00:00+00:00
- Products: 
- Tags: release, feature
- URL: https://standarddb.com/zed/zed-ships-split-diff-view-as-default-in-v0-224-maintains-alignment-at-scale--jjbhlpvq
- Original: https://zed.dev/blog/split-diffs

## Split Diff View Now Default

Zed v0.224 introduces split diffs as the new default view for project and branch diffs. The split view displays the base version of files on the left and your working copy on the right, with both sides scrolling in sync. This was the most-requested feature following Zed's Git integration launch in March 2025.

## Technical Architecture

Unlike traditional file-by-file diff implementations, Zed's split diffs are built on the editor's **multibuffer** system—a unified, editable surface spanning multiple files. This architecture enables seamless cursor movement across file boundaries and text selection spanning multiple files, but it introduced unique technical challenges:

- **Alignment complexity**: A single misalignment in the split view cascades through the entire multibuffer, affecting file headers and content placement across all changed files
- **Scalability requirements**: The implementation had to handle projects with thousands of changed files without performance degradation

## Implementation Details

The team modeled the left-side multibuffer by using the original file versions (already computed for diffs) and highlighting deleted hunks in red. To keep corresponding lines vertically aligned, they leveraged Zed's **block map**—the same component powering diagnostics and inline assistance—to insert spacers (visual padding rendered as subtle checkerboard patterns) between lines.

## Performance Optimizations

Development surfaced unexpected wins across the editor:

- Fixed inefficiencies in the block map discovered during view switching, improving project search performance
- Corrected macOS process spawning API usage (`posix_spawn` instead of `fork`/`exec`), reducing main-thread hangs from git blame and external processes
- Split-diff-specific optimizations for initial view construction and incremental diff updates

## Getting Started

Split diff is now the default behavior, but users can revert to unified diffs via **Diff View Style** settings. The team continues optimizing performance in upcoming releases.

---

### GitHub Desktop 3.5.5 adds Git hooks support
- Published: 2026-02-18T00:00:00+00:00
- Products: GitHub
- Tags: release, feature, bugfix
- URL: https://standarddb.com/github/github-desktop-3-5-5-brings-native-git-hooks-support-with-environment-variable--cnwpmwkd
- Original: https://github.blog/changelog/2026-02-18-github-desktop-3-5-5-adds-git-hooks-support

### Git Hooks Finally Work Properly

GitHub Desktop has struggled with Git hooks compatibility for years, particularly on Windows where the application ships with an embedded Git installation isolated from your shell environment. This prevented hooks from accessing critical dependencies like Node version managers (nvm), environment-specific configuration, or tools not bundled with Desktop. Additionally, hook failures were indistinguishable from other Git errors, and terminal escape codes rendered as garbled text.

### What's New in 3.5.5

Hooks now have full access to your shell environment by sourcing configuration files like `.bash_profile` or `.zshrc`. This resolves the majority of compatibility issues with version managers and environment-dependent scripts. Key improvements include:

- **Environment variable support**: Hooks can now access all variables from your shell configuration
- **Real-time output with proper formatting**: Terminal colors and formatting now display correctly
- **Hook control**: Enable/disable hooks in Settings > Git > Hooks
- **Bypass options**: Skip commit hooks preemptively via the gear icon, or bypass a failed hook without aborting your commit

### Additional Improvements

Beyond hooks support, version 3.5.5 includes:

- Warp terminal support for Windows
- Open repositories in different editors without changing default preferences
- Quick GitHub viewing for branches via right-click menu
- Fixed crashes when Git output contains emoji or multibyte characters
- Resolved repository state issues when switching branches with submodules
- Copilot-authored commits now display the Copilot avatar

### Getting Started

Automatic updates are rolling out progressively, or download v3.5.5 directly from the [GitHub Desktop releases page](https://github.com/desktop/desktop/releases/tag/release-3.5.5). Report issues or provide feedback on the [GitHub Desktop issue tracker](https://github.com/desktop/desktop/issues/new/choose).

---

### Secret scanning improvements to extended metadata checks
- Published: 2026-02-18T00:00:00+00:00
- Products: GitHub
- Tags: feature, security, api
- URL: https://standarddb.com/github/github-secret-scanning-adds-extended-metadata-checks-configurable-at--yyrtdvsk
- Original: https://github.blog/changelog/2026-02-18-secret-scanning-improvements-to-extended-metadata-checks

## Extended Metadata Checks Now Available

GitHub secret scanning is expanding its capabilities with improved support for extended metadata checks in security configurations. This enhancement makes it substantially easier to enable and manage the feature at scale across your organization.

### What Extended Metadata Checks Provide

Extended metadata checks enhance secret scanning alerts by displaying rich context about leaked secrets:

- **Secret owner information**: Name, email, and identifier of the secret owner
- **Temporal data**: Secret creation and expiry dates
- **Organizational context**: Project or organization information associated with the secret
- **Provider-specific details**: Additional metadata available from the secret provider (e.g., OpenAI keys display owner and organization information)

These metadata additions expand on existing validity checks to provide development and security teams with actionable context for faster triage, remediation assessment, and prioritization.

### What's Changing

Extended metadata checks are now available to **Enterprise Cloud customers** with secret scanning and validity checks enabled. Key improvements include:

- **Configuration flexibility**: Enable or disable extended metadata checks at organization and enterprise levels using security configurations
- **Automatic enablement**: Repositories with validity checks already enabled via security configurations will have metadata checks automatically enabled
- **Audit trail**: Track feature enablement status by monitoring enterprise or organization audit logs

### Important Notes

The availability of metadata depends on several factors:

- The specific secret provider
- The type of token or credential
- The individual secret itself

GitHub makes a best-effort approach to display all available metadata, but not every secret will contain all metadata fields.

### Next Steps

Learn more about [securing your repositories with secret scanning](https://docs.github.com/enterprise-cloud@latest/code-security/concepts/secret-security/about-secret-scanning) in the GitHub documentation, or [share feedback](https://github.com/orgs/community/discussions/177054) about this feature.

---

### Analytics - New cfWorker metric in Server-Timing header
- Published: 2026-02-18T00:00:00+00:00
- Products: Cloudflare Workers, Cloudflare
- Tags: feature, api, performance, sdk
- URL: https://standarddb.com/cloudflare/cloudflare-adds-cfworker-metric-to-server-timing-header-for-performance--xvtjtpfh
- Original: https://developers.cloudflare.com/changelog/post/2026-02-18-cfworker-server-timing

## Overview

The Server-Timing header now includes a dedicated `cfWorker` metric that breaks down execution time spent in Cloudflare Workers. Previously, Worker execution time was bundled into the `edge` metric, making it difficult to pinpoint performance bottlenecks between edge processing and upstream dependencies.

## What Changed

The metric breakdown now clearly separates:

- **edge**: Total time on Cloudflare edge (including Worker execution)
- **origin**: Time spent fetching from origin server
- **cfWorker**: Time spent in Worker execution and subrequests (excluding origin fetch)

Example response header:
```
Server-Timing: cdn-cache; desc=DYNAMIC, edge; dur=20, origin; dur=100, cfWorker; dur=7
```

## Availability and Enablement

The `cfWorker` metric is **enabled by default** if you have Real User Monitoring (RUM) enabled. Users without RUM can enable it via Rules configuration in the Cloudflare dashboard.

## Benefits for Developers

- **Performance debugging**: Quickly determine whether latency stems from Worker code, external API calls within Workers, or slow origins
- **Optimization targeting**: Identify which component of the request path requires optimization
- **Real User Monitoring**: Access detailed timing breakdowns directly from response headers for client-side analytics and monitoring

The metric is particularly valuable for developers running complex Workers with multiple subrequests or external API integrations.

---

### Introducing Weaviate Agent Skills
- Published: 2026-02-18T00:00:00+00:00
- Products: 
- Tags: feature, integration, sdk, open-source, platform
- URL: https://standarddb.com/weaviate/weaviate-launches-agent-skills-library-to-guide-ai-coding-assistants-on-vector--vtxbmdkq
- Original: https://weaviate.io/blog/weaviate-agent-skills

## Overview

Weaviate has introduced Agent Skills, an open-source library designed to improve how AI coding assistants interact with Weaviate's vector database infrastructure. The library bridges the gap between agent-driven development ("vibe coding") and specialized tooling, ensuring that Claude Code, Cursor, GitHub Copilot, and other agents can correctly implement Weaviate operations without hallucinating legacy syntax or misunderstanding API parameters.

## What's Included

The repository is organized into two main tiers:

- **Weaviate Skills** (`/skills/weaviate`): Focused scripts for cluster management (schema inspection, collection creation), data lifecycle operations (CSV/JSON/JSONL imports), agentic search via the Query Agent, and advanced retrieval methods (hybrid, semantic, keyword search with fine-tuned parameters)
- **Cookbooks** (`/skills/weaviate-cookbooks/`): End-to-end project blueprints including Query Agent chatbots (FastAPI + optional Next.js), PDF retrieval with multivector embeddings, and basic to advanced RAG pipelines

## Built-in Commands

The library provides six commands for Claude Code Plugin users:

- **Ask**: Query with AI-generated answers and source citations
- **Collections**: List collections or retrieve collection schemas
- **Explore**: Inspect collection data with property metrics and sample objects
- **Fetch**: Retrieve and filter objects by ID or custom criteria
- **Query**: Natural language search via Query Agent
- **Search**: Hybrid, semantic, or keyword search with raw results

## Getting Started

Developers can access the repository at [https://github.com/weaviate/agent-skills](https://github.com/weaviate/agent-skills). The `/weaviate:quickstart` command provides a guided setup, while `/weaviate:data` supplies example datasets to accelerate development. Simple natural language prompts like "Create a Weaviate collection for my JSON data" or "Build a chatbot using the Query Agent" enable agents to generate correct, production-ready code on the first attempt.

---

### A home for everything in Library
- Published: 2026-02-18T00:00:00+00:00
- Products: 
- Tags: feature, platform
- URL: https://standarddb.com/notion/notion-adds-library-full-page-view-to-organize-workspace-content-as-sidebar--qgwswppp
- Original: https://www.notion.so/releases/2026-02-18

## Library: A New Organization Solution

As Notion workspaces grow, sidebar navigation can become cluttered and difficult to manage. **Library addresses this problem by providing a dedicated full-page view** for organizing and accessing frequently-used content types.

## Key Features

Library organizes your workspace content into five distinct tabs:

- **Teamspaces** - Access shared team areas
- **Recents** - Quickly jump to recently viewed pages
- **Favorites** - Find your starred content
- **Private pages** - View your personal workspace items
- **Shared pages** - See pages shared with you by others

## Benefits for Users

The new view enables a cleaner, more efficient workspace navigation experience. Users can **customize their sidebar to display only their most-used items daily**, while less-frequently accessed content is available in Library without cluttering the primary navigation. This separation improves both usability and visual clarity, especially for growing teams and complex workspaces.

## How to Get Started

To use Library, users can access it as a new tab in their workspace interface and begin organizing their content across the five categories. Customizing sidebar visibility is straightforward, allowing users to optimize their workspace navigation to match their personal workflow.

---

### CLI Improvements and Mermaid ASCII Diagrams
- Published: 2026-02-18T00:00:00+00:00
- Products: 
- Tags: release, feature, platform
- URL: https://standarddb.com/cursor/cursor-cli-gains-plan-handoff-to-cloud-and-inline-mermaid-diagram-rendering--rryfksdf
- Original: https://cursor.com/changelog/cli-feb-18-2026

## Plan Mode Improvements

The CLI now displays a persistent decision menu when plans are generated, allowing developers to choose between building locally or executing in the cloud. The `/plan` command provides quick access to the current plan and its action menu, while new keyboard shortcuts streamline navigation:

- **Arrow keys** to navigate options
- **Enter** to execute the selected option  
- **Shift+Enter** as a shortcut for "Build in cloud"

## Mermaid ASCII Diagrams

Mermaid code blocks now render directly as ASCII diagrams within CLI conversations, supporting multiple diagram types:

- Flowcharts
- Sequence diagrams
- State machines
- Class diagrams
- ER diagrams

Users can press **Ctrl+O** to toggle between the rendered ASCII visualization and the original Mermaid source code, making it easy to reference both representations.

## Additional Improvements

Beyond the headline features, this release includes 42 improvements across tooling, UX, and reliability, with updates spread across general functionality, tools integration, user experience, and bug fixes.

---

### Introducing BigLaw Bench: Global
- Published: 2026-02-18T00:00:00+00:00
- Products: 
- Tags: release, model, platform, feature
- URL: https://standarddb.com/harvey/harvey-launches-biglaw-bench-global-expanding-ai-benchmarking-to-uk-australia--wtrtbtct
- Original: https://www.harvey.ai/blog/introducing-big-law-bench-global

## Expanding Global Benchmarking for Legal AI

Harvey has introduced **BigLaw Bench: Global (BLB: Global)**, an extension of its existing BigLaw Bench framework that measures how well AI models perform on core legal tasks across different jurisdictions. This new dataset represents a significant expansion of Harvey's benchmarking capabilities, more than doubling the size of the public-facing BigLaw Bench.

## Built for Global Legal Practice

BLB: Global was developed in collaboration with Mercor, a leading expert data company, to capture the nuances of legal work across different jurisdictions. The initial release focuses on three markets:

- **United Kingdom**
- **Australia**
- **Spain**

The framework is designed to help foundation models understand and execute core AI workflows—such as drafting, document analysis, and legal research—with accuracy and consistency while respecting local legal norms and practices.

## Addressing Localization Challenges

While leading AI models now achieve around 90% performance on the original BigLaw Bench tasks, Harvey's research identified a critical gap: model performance often degrades when applied to localized legal contexts. BLB: Global addresses this by benchmarking six major task categories that customers depend on, including:

- **Drafting:** Production of legal documents, memos, and contracts
- Additional core legal workflows (details partially cut off in original content)

## Implications for Legal Teams

This benchmark extension enables Harvey to measure and improve how well AI models localize across jurisdictions, helping ensure customers worldwide can rely on accurate, consistent AI assistance aligned with their local legal requirements. The framework provides a foundation for expanding to additional jurisdictions over time.

---

### llama-index-core [0.14.15]
- Published: 2026-02-18T00:00:00+00:00
- Products: 
- Tags: release, feature, api, bugfix, integration
- URL: https://standarddb.com/llamaindex/llamaindex-0-14-15-adds-multimodal-support-and-recursive-llm-types--hlkyyrgm
- Original: https://github.com/run-llama/llama_index/blob/main/CHANGELOG.md

## Multimodal Type Support

LlamaIndex core 0.14.15 significantly expands multimodal capabilities with support for basic operations on multimodal types. This includes new multimodal template variable formatting and multimodal prompt templates, enabling developers to work with non-text content more seamlessly.

## Recursive LLM Types

The release introduces support for recursive LLM type definitions, allowing for more complex type hierarchies and nested structures in LLM interactions.

## Enhanced Prompt Templating

Three new features improve prompt handling:
- **Multimodal template variable formatting** for flexible content insertion
- **Multimodal prompt templates** for structured multimodal interactions
- **Multimodal chat prompt helper** to simplify building multimodal conversations

## Robustness Improvements

- **BaseExtractor retry and error handling** (#20693) adds resilience to data extraction operations
- **Memory fix** ensures at least one message/content block is returned, preventing empty response issues
- **TextNode cleanup** removes redundant metadata_separator field

## Integration Updates

Beyond core changes, this release includes updates across the ecosystem:
- **Anthropic LLM**: Added Claude Sonnet 4.6 support
- **Bedrock Converse**: Added Sonnet 4.6 and fixed thinking_delta population
- **Mistral AI**: Updated to latest Azure SDK
- **OCI DataScience**: New streaming endpoint support
- **OpenTelemetry**: Improved data serialization with custom span processor support

---

### v0.5.92
- Published: 2026-02-17T23:58:07+00:00
- Products: Sim Studio
- Tags: release, feature, integration, bugfix, performance
- URL: https://standarddb.com/sim/sim-studio-v0-5-92-adds-pipedrive-sort-order-and-improves-pagination-across--ztxcrzqg
- Original: https://github.com/simstudioai/sim/releases/tag/v0.5.92

## New Features

Sim Studio v0.5.92 brings several improvements to integration capabilities and platform stability:

- **Pipedrive Enhancement**: Added sort order support to Pipedrive endpoints, alongside a turborepo upgrade to improve build performance.
- **Pagination Expansion**: Extended pagination support to remaining integrations that support it, providing better handling of large datasets.
- **Short Link Rewrite**: Implemented Beluga short link rewrite functionality for hosted campaigns.

## Code Quality & Type Safety

The release focuses on improving code maintainability:

- **Type Safety**: Replaced `@ts-ignore` comments with a properly typed `ProviderError` class, reducing technical debt and improving error handling across the codebase.
- **Linting Improvements**: Fixed react-doctor errors and warnings to maintain code quality standards.
- **UI Improvements**: Enhanced copilot scrolling stickiness for better user experience.

## Bug Fixes

- Fixed pagination by adding missing `next_page` fields to response interfaces and operator comments, ensuring consistent pagination behavior across endpoints.

This incremental release continues Sim Studio's focus on expanding integration capabilities and improving code quality.

---

### NVIDIA Nemotron 2 Nano 9B Japanese: 日本のソブリンAIを支える最先端小規模言語モデル
- Published: 2026-02-17T23:28:52+00:00
- Products: 
- Tags: release, model, feature
- URL: https://standarddb.com/huggingface/nvidia-releases-nemotron-nano-9b-v2-japanese-tops-sub-10b-category-on-nejumi--lcdjglkl
- Original: https://huggingface.co/blog/nvidia/nemotron-nano-9b-v2-japanese-ja

## Overview

NVIDIA has released **Nemotron-Nano-9B-v2-Japanese**, a Japanese-specialized variant of its efficient small language model. The model achieves top performance on the Nejumi Leaderboard 4 in the sub-10B parameter category, addressing a critical gap in the Japanese enterprise AI market where few models combine both advanced Japanese proficiency and agentic task execution capabilities.

## Key Features & Capabilities

The model maintains the proven architecture of the English-optimized Nemotron-Nano-9B-v2 while adding specialized Japanese language capabilities. Notable capabilities include:

- **Advanced Japanese language understanding** with strong reasoning and knowledge retention
- **Robust tool-calling and agent abilities** for function invocation and multi-step workflows
- **Instruction-following** and alignment across multiple dimensions
- **Efficient parameter count** (9B) enabling on-premises deployment without significant infrastructure investment

## Training Approach

The development leverages two foundational pillars:

**Nemotron-Nano-9B-v2 Architecture**: A proven baseline with excellent size-to-performance ratio, adapted from the English-trained model.

**Nemotron-Personas-Japan Dataset**: An open-source (CC BY 4.0) dataset of 6M demographically and geographically representative Japanese personas used as seed data for synthetic data generation. This culturally-grounded approach ensures training data reflects real-world Japanese diversity and use cases, particularly for tool-calling scenarios.

The training pipeline combines:
- Continued pretraining on Japanese OSS corpora (Wikipedia, fineweb-2-ja, aozorabunko, sip3-ja-general-web-corpus)
- Synthetic data generation (SDG) using Nemotron-Personas-Japan as seed set
- Supervised fine-tuning (SFT) with tool-calling and instruction-following data
- Training uses **Megatron-LM** and **NeMo Curator** tools

## Enterprise Relevance

The model addresses specific Japanese enterprise needs:

- **On-premises deployment**: Sub-10B size enables private network deployment for organizations handling sensitive data
- **Customization efficiency**: Strong base capabilities reduce fine-tuning cycles compared to foundation models
- **Agent development**: Proven agent architecture supports rapid prototyping of multi-agent systems without large-model overhead

## Availability

The model is available on Hugging Face at `nvidia/NVIDIA-Nemotron-Nano-9B-v2-Japanese`. Related datasets including Nemotron-Personas for other regions (US, India, Singapore, Brazil) are also available, enabling similar customization approaches across markets.

---

### v0.16.3-rc0
- Published: 2026-02-17T21:58:49+00:00
- Products: 
- Tags: 
- URL: https://standarddb.com/ollama/v0-16-3-rc0--gjxgflqc
- Original: https://github.com/ollama/ollama/releases/tag/v0.16.3-rc0



---

### Neon Is a Cursor Plugin
- Published: 2026-02-17T18:58:33+00:00
- Products: 
- Tags: integration, feature, platform
- URL: https://standarddb.com/neon/neon-launches-cursor-plugin-for-ai-assisted-postgres-workflows--bzmdkddl
- Original: https://neon.com/blog/neon-is-a-cursor-plugin

## Neon Plugin Now Available in Cursor Marketplace

Neon has launched an official plugin for [Cursor](https://cursor.com/), the AI-powered code editor, available immediately in the Cursor Marketplace. The plugin gives Cursor live access to your Neon organization, enabling AI-assisted database management directly within the editor.

### What's Included

The Neon plugin bundles two core components:

- **Neon MCP Server** — Provides live access to your Neon organization and databases
- **Neon Skill** — Defines structured workflows and best practices for database operations with Neon

### Getting Started

Installing the plugin is straightforward:

1. Visit the [Cursor Marketplace](https://cursor.com/marketplace/neon)
2. Click "Add to Cursor"
3. The plugin becomes available across all Cursor surfaces (desktop, CLI, web, and mobile)

### Why This Matters

With this integration, developers can now:

- **Move beyond code generation** — Cursor can assist with database schema design, migrations, and optimization
- **Reduce context switching** — Manage Postgres workflows without leaving your editor
- **Leverage AI for database tasks** — Get AI-assisted help with real infrastructure work, not just code

The Neon plugin joins Cursor's initial marketplace launch alongside integrations for Linear, Notion, Figma, Vercel, Stripe, Sentry, and Databricks, creating a comprehensive development stack directly accessible from the AI editor.

---

### Vercel Sandbox snapshots now allow custom retention periods
- Published: 2026-02-17T13:00:00+00:00
- Products: Vercel
- Tags: feature, sdk, api
- URL: https://standarddb.com/vercel/vercel-sandbox-snapshots-gain-configurable-retention-raising-default-limit-from--czllzfgn
- Original: https://vercel.com/changelog/vercel-sandbox-snapshots-now-allow-custom-retention-periods

## Configurable Snapshot Retention

Vercel Sandbox snapshots now support flexible expiration settings, eliminating the previous hard limit of 7 days. This gives developers greater control over snapshot lifecycle management and aligns with longer-term development workflows.

## Key Changes

- **Expiration range**: Configure snapshots to expire between 1 day and infinity
- **New default**: Snapshots now default to 30 days instead of 7 days
- **API support**: Set expiration via the SDK using millisecond values (e.g., `expiration: ms('14d')`)
- **CLI support**: Configure retention from the command line with `--expiration` flag

## Implementation

**SDK usage:**
```typescript
const sandbox = Sandbox.create();
sandbox.snapshot({ expiration: ms('1d') })
```

**CLI usage:**
```bash
# 14-day expiration
sandbox snapshot sb_1234567890 --stop --expiration 14d

# Never expire
sandbox snapshot sb_1234567890 --stop --expiration 0
```

## What Developers Need to Know

This change is backward compatible—existing snapshot workflows continue to work with the new 30-day default. Developers relying on the previous 7-day retention should review and update their snapshot expiration settings if longer retention is desired. The flexibility to set expiration to infinity is useful for long-lived development environments and reference snapshots.

---

### Automatic build fix suggestions with Vercel Agent
- Published: 2026-02-17T13:00:00+00:00
- Products: Vercel
- Tags: feature, platform, integration
- URL: https://standarddb.com/vercel/vercel-agent-now-suggests-automatic-build-fixes-in-pull-requests--xvqgzmbc
- Original: https://vercel.com/changelog/automatic-build-fix-suggestions-with-vercel-agent

## Automatic Build Error Detection and Code Suggestions

Vercel Agent now includes intelligent build failure detection and automatic fix suggestions. When reviewing pull requests, the Agent scans your deployments for build errors and generates targeted code fix recommendations based on your actual code and build logs.

## Where Fixes Appear

The automatic suggestions are available in two places:

- **GitHub Pull Requests**: Code suggestions appear directly in your PR review, allowing you to review and approve fixes before merging
- **Vercel Dashboard**: Build fix suggestions are also available in the Dashboard when errors are detected, with the option to create a GitHub PR with the suggested changes

## Getting Started

To start using Vercel Agent with automatic build fix suggestions, access the [Agent dashboard](https://vercel.com/~/agent) or consult the [documentation](https://vercel.com/docs/agent) for detailed setup instructions. The feature is available for all Vercel users and integrates seamlessly with existing GitHub workflows.

---

### Improved streaming runtime logs exports
- Published: 2026-02-17T13:00:00+00:00
- Products: Vercel
- Tags: feature, platform
- URL: https://standarddb.com/vercel/vercel-streaming-runtime-logs-exports-now-enable-background-downloads-without--cxymzlsm
- Original: https://vercel.com/changelog/improved-streaming-runtime-logs-exports

## Streaming Downloads Without Buffering

Runtime logs exports now stream directly to your browser, eliminating the need to wait for large files to buffer. Downloads start immediately and run in the background, allowing you to continue working in the Vercel dashboard while the export completes.

## New Export Options

Two new filtering options are now available:
- **Export visible data**: Download exactly what's currently displayed on your screen
- **Export filtered results**: Export all requests matching your active search filters

## Updated Export Limits

All plans can export up to **10,000 requests per export**, while [Observability Plus](https://vercel.com/docs/observability/observability-plus) subscribers can export up to **100,000 requests** – a significant increase from previous limits.

## Consistency Improvements

Exported log data is now indexed by request to ensure consistency with the Runtime Logs dashboard interface. Export limits are applied per request to guarantee that exported data matches the filtered requests shown on the dashboard.

---

### Automated security audits now available for skills.sh
- Published: 2026-02-17T13:00:00+00:00
- Products: Vercel
- Tags: feature, security, platform
- URL: https://standarddb.com/vercel/vercel-adds-automated-security-audits-to-skills-sh-for-60-000-packages--qbcfgbwd
- Original: https://vercel.com/changelog/automated-security-audits-now-available-for-skills-sh

## Security Audits Now Live on skills.sh

Vercel has introduced **automated security audits** for skills on skills.sh, leveraging partnerships with Gen, Socket, and Snyk to provide independent security validation at scale. These audits cover over 60,000 skills and counting, helping developers make informed decisions when integrating third-party skills into their projects.

## Key Features and Improvements

The new security audit system includes:

- **Transparent Results**: Security audit reports appear publicly on each skill's detail page, allowing developers to review findings before installation.
- **Automatic Protection**: Skills flagged as malicious are automatically hidden from the leaderboard and search results. Direct navigation to flagged skills displays a warning before installation.
- **Installation Validation**: Starting with `skills@1.4.0`, the skills CLI displays audit results and risk levels during the installation process.

## Developer Impact

Developers using skills.sh can now confidently browse and integrate skills with clear visibility into security posture. The automated audits reduce the burden of manual security reviews while maintaining ecosystem trust through transparent reporting and aggressive filtering of malicious packages.

---

### Claude Sonnet 4.6 is live on AI Gateway
- Published: 2026-02-17T13:00:00+00:00
- Products: Vercel AI SDK, Vercel
- Tags: release, model, api, integration, feature
- URL: https://standarddb.com/vercel/claude-sonnet-4-6-now-available-on-vercel-ai-gateway-with-1m-token-context--cybwtcmw
- Original: https://vercel.com/changelog/claude-sonnet-4-6-is-live-on-ai-gateway

## Claude Sonnet 4.6 on AI Gateway

Anthropic's Claude Sonnet 4.6 is now live on Vercel's AI Gateway, expanding the available models for developers building with Vercel's unified API layer. This release brings **Opus-level intelligence** to a faster, more cost-effective tier, with substantial improvements across agentic coding, code review, frontend UI quality, and computer use accuracy.

### Key Capabilities

The model includes several advanced features:
- **Agentic execution**: Proactively executes tasks, delegates to subagents, and parallelizes tool calls
- **MCP support**: Enables scaled tool use through Model Context Protocol integration
- **Hybrid reasoning**: Delivers both near-instant responses and extended thinking within the same model
- **1M token context window**: Supports handling of very large documents and codebases
- **Adaptive thinking**: Control reasoning depth with configurable `effort` levels and `adaptive` thinking type

### Integration with Vercel AI SDK

Developers can access Sonnet 4.6 by setting the model identifier to `anthropic/claude-sonnet-4.6` in the AI SDK. The model supports configuration options for reasoning effort and thinking type:

```tsx
const result = streamText({
  model: 'anthropic/claude-sonnet-4.6',
  prompt: 'Your task here...',
  providerOptions: {
    anthropic: {
      effort: 'medium',
      thinking: { type: 'adaptive' },
    },
  },
});
```

### Vercel AI Gateway Benefits

Beyond model availability, AI Gateway provides developers with:
- **Unified API**: Single interface for calling multiple models
- **Observability**: Built-in tracking of usage, costs, and performance
- **Reliability**: Automatic retries, failover, and intelligent provider routing for higher uptime than direct provider access
- **Bring Your Own Key**: Option to supply your own API credentials
- **Performance optimizations**: Configured at deployment time

Developers can explore Sonnet 4.6 immediately through the [model playground](https://vercel.com/ai-gateway/models/claude-sonnet-4.6) or consult the [AI Gateway model leaderboard](https://vercel.com/ai-gateway/leaderboards) for performance comparisons.

---

### Recraft V4 on AI Gateway
- Published: 2026-02-17T13:00:00+00:00
- Products: Vercel AI SDK, Vercel
- Tags: release, feature, model, integration, api
- URL: https://standarddb.com/vercel/recraft-v4-image-model-now-available-on-vercel-ai-gateway--qddtqjvt
- Original: https://vercel.com/changelog/recraft-v4-on-ai-gateway

## Recraft V4 Now Available on AI Gateway

Recraft V4, a professional-grade text-to-image model, is now integrated into Vercel's AI Gateway. Developed with input from working designers, this model addresses real-world needs for design and marketing use cases.

### Key Improvements

The V4 model delivers significant improvements in photorealism, featuring:
- **Realistic skin rendering** and natural textures with fewer synthetic artifacts
- **Clean lighting and varied composition** for professional-quality output
- **Better character generation** for illustration work with less predictable, more original color palettes

### Two Model Variants

**V4** - The standard version optimized for speed and cost efficiency, ideal for everyday design work and rapid iteration cycles.

**V4 Pro** - The premium variant that generates higher-resolution images suitable for print-ready assets and large-scale deployment.

### Integration with AI SDK

Developers can integrate Recraft V4 immediately using Vercel's AI SDK by specifying the model parameter:

```tsx
const result = await generateImage({
  model: 'recraft/recraft-v4',
  prompt: 'Product photo of a ceramic coffee mug on a wooden table, morning light, shallow depth of field.'
});
```

### AI Gateway Benefits

Beyond model access, AI Gateway provides unified API infrastructure including usage tracking, cost monitoring, intelligent retry logic, provider failover, and built-in observability features. The service also supports bring-your-own-key authentication for leveraging existing provider accounts.

---

### Claude Sonnet 4.6 is now available in Windsurf
- Published: 2026-02-17T12:00:00+00:00
- Products: 
- Tags: feature, model, release, integration
- URL: https://standarddb.com/windsurf/claude-sonnet-4-6-launches-in-windsurf-with-2-3x-promotional-pricing--djfdfkjz
- Original: https://windsurf.com/blog/sonnet-4.6

## Claude Sonnet 4.6 Now Available in Windsurf

Claude Sonnet 4.6 has launched in Windsurf with limited-time promotional pricing for self-serve users. This marks the availability of Anthropic's latest reasoning model in Exafunction's AI code editor.

### Pricing & Availability

The promotional pricing structure offers:
- **Without extended thinking:** 2x credits
- **With extended thinking:** 3x credits

Claude Sonnet 4.6 is also available in Arena Mode's Frontier and Hybrid battle groups for users comparing different models.

### Key Capabilities

According to statements from both CEOs, Sonnet 4.6 delivers significant improvements:
- **Frontier-level reasoning** in a smaller, more cost-effective form factor
- **Viable alternative for heavy Opus users** seeking better cost-efficiency
- **Improved bug detection** that meaningfully closes the gap with Opus, enabling parallel reviewer execution and broader bug detection without cost increases

### Getting Started

Users can download the latest version of Windsurf at [windsurf.com/download](https://windsurf.com/download) to access Claude Sonnet 4.6 immediately.

---

### v0.16.2
- Published: 2026-02-17T00:15:52+00:00
- Products: 
- Tags: release, bugfix, feature
- URL: https://standarddb.com/ollama/ollama-v0-16-2-adds-web-search-for-claude-cloud-models-fixes-image-generation--jhqdzfzv
- Original: https://github.com/ollama/ollama/releases/tag/v0.16.2

## Key Changes

**Cloud Model Enhancements**
- `ollama launch claude` now supports web search when using `:cloud` models, enabling Claude to retrieve and reference live web information during conversations
- New app setting makes it easier to disable cloud models for environments where data privacy is critical and information cannot leave the local machine
- For Linux deployments or manual `ollama serve` configurations, set `OLLAMA_NO_CLOUD=1` environment variable to disable cloud model access

**Bug Fixes**
- Fixed rendering issue that occurred when running Ollama in PowerShell environments
- Resolved regression where experimental image generation models would fail to execute in versions 0.16.0 and 0.16.1

## What Developers Need to Know

This release focuses on improving usability and fixing critical regressions. Users relying on image generation capabilities should upgrade immediately. Teams handling sensitive data can now easily isolate their Ollama instances from cloud-based models, enhancing compliance with data residency requirements.

## Action Items

- Upgrade to v0.16.2 if using experimental image generation features
- Consider enabling `OLLAMA_NO_CLOUD=1` for deployments handling sensitive data
- Test cloud model web search capabilities with Claude models in `:cloud` mode

---

### Required reviewer rule is now generally available
- Published: 2026-02-17T00:00:00+00:00
- Products: GitHub
- Tags: release, feature, api
- URL: https://standarddb.com/github/github-s-required-reviewer-rule-exits-preview-with-negation-pattern-support--lklqtqhr
- Original: https://github.blog/changelog/2026-02-17-required-reviewer-rule-is-now-generally-available

## Required Reviewer Rule Now GA

GitHub's **required reviewer rule** for repository rulesets has moved from public preview to general availability. This feature enables organizations and enterprises to enforce granular approval policies, ensuring that specific teams must review and approve changes to designated branches and files before merging.

## Key Capabilities

With the required reviewer rule, teams can:

- **Enforce approval thresholds** — Require a specific number of approvals from designated teams before merging into protected branches
- **Precise file targeting** — Use pattern matching to apply policies to specific files and folders
- **Exclude paths with negation** — NEW since preview: Use `!` syntax (like `.gitignore`) to exclude specific files or folders from review requirements
- **Enterprise-wide scaling** — Apply review policies consistently across repositories, organizations, or entire enterprises

## Complementing CODEOWNERS

The required reviewer rule focuses on **policy enforcement** rather than ownership definition. It augments but does not replace CODEOWNERS files, which remain the best way to manage code ownership and support individual reviewers.

For example, organizations can now:
- Require all `*.sql` changes to have review approval from the data platform team
- Mandate two security team reviews for authentication file changes before merging to `main`
- Define flexible rules that scale consistently across teams and repositories

## Getting Started

Documentation for repository rulesets is available in [GitHub's official docs](https://docs.github.com/repositories/configuring-branches-and-merges-in-your-repository/managing-rulesets/about-rulesets). Teams can join discussions and share feedback in [GitHub Community](https://github.com/orgs/community/discussions/categories/repositories).

---

### Access - Streamlined clientless access for private applications
- Published: 2026-02-17T00:00:00+00:00
- Products: Cloudflare
- Tags: feature, api
- URL: https://standarddb.com/cloudflare/cloudflare-access-streamlines-clientless-app-access-with-new-policy-setting--pbydnmmm
- Original: https://developers.cloudflare.com/changelog/2026-02-17-clientless-access-for-private-apps

## Simplified Clientless Access Management

Cloudflare Access has introduced a new **Allow clientless access** setting that streamlines the process of enabling browser-based access to private self-hosted applications. This feature eliminates the need for the previous workaround of creating separate bookmark applications.

## What Changed

Previously, organizations had to:
- Create a separate [bookmark application](https://developers.cloudflare.com/cloudflare-one/access-controls/applications/bookmarks/)
- Point it to a prefixed [Clientless Web Isolation](https://developers.cloudflare.com/cloudflare-one/remote-browser-isolation/setup/clientless-browser-isolation/) URL (e.g., `https://<your-teamname>.cloudflareaccess.com/browser/https://10.0.0.1/`)
- Accept that this bookmark would be visible to all App Launcher users, regardless of access permissions

## New Workflow

With the update, administrators can now:
- **Enable clientless access directly** within their private self-hosted application settings
- **Control visibility** so only users who pass the Access application policies see the clientless access tile in their App Launcher
- **Maintain security** by requiring users to have remote browser permissions to open the link

This change reduces configuration complexity and improves the security posture by tying clientless access permissions directly to the underlying application's access policies.

---

### Introducing Claude Sonnet 4.6
- Published: 2026-02-17T00:00:00+00:00
- Products: Claude Code, Claude, Anthropic
- Tags: release, model, feature, api, performance
- URL: https://standarddb.com/anthropic/anthropic-releases-claude-sonnet-4-6-with-70-preference-over-predecessor-in--jcmzrxnw
- Original: https://www.anthropic.com/news/claude-sonnet-4-6

## A Full Upgrade Across Core Capabilities

Anthropic has launched Claude Sonnet 4.6, the latest iteration of its mid-tier model that balances capability with cost-efficiency. The model represents a significant jump in performance across coding, computer use, long-context reasoning, agent planning, knowledge work, and design tasks. It's now the default model for Free and Pro plan users on claude.ai and Claude Cowork.

## Coding and Instruction Following

Early testing shows developers strongly prefer Sonnet 4.6 over Sonnet 4.5 (70% of the time) and even over the frontier Opus 4.5 model (59% of the time). Users report that Sonnet 4.6:

- More effectively reads context before modifying code
- Consolidates shared logic rather than duplicating it
- Shows significantly less overengineering and "laziness"
- Is substantially better at instruction following
- Produces fewer false claims of success and hallucinations
- Delivers more consistent follow-through on multi-step tasks

This performance-at-cost profile means organizations can now accomplish work previously requiring expensive Opus-class models using the more economical Sonnet tier.

## Computer Use and Extended Context

Sonnet 4.6 marks a major improvement in computer use capabilities, advancing on the OSWorld benchmark from 20.1% (Sonnet 4.5) to higher performance levels. The model can now handle human-level tasks like navigating complex spreadsheets and filling out multi-step web forms before coordinating across multiple browser tabs—all without special APIs or connectors.

The 1M token context window (in beta) is a practical upgrade, allowing the model to reason effectively across entire codebases, lengthy contracts, or dozens of research papers in a single request. Early testing via the Vending-Bench Arena evaluation demonstrated Sonnet 4.6's sophisticated long-horizon planning: the model invested heavily in capacity early on, then pivoted to profitability to outcompete other AI models.

## Safety and Pricing

Anthropic has completed extensive safety evaluations showing Sonnet 4.6 is as safe as or safer than recent Claude models. The model demonstrates strong resistance to prompt injection attacks compared to Sonnet 4.5 and performs similarly to Opus 4.6. Pricing remains unchanged from Sonnet 4.5: $3 per million input tokens and $15 per million output tokens on the API.

Developers can start using Sonnet 4.6 immediately on claude.ai and via the API at the standard Sonnet pricing tier.

---

### Announcing the Slack MCP server and Real-time Search API
- Published: 2026-02-17T00:00:00+00:00
- Products: 
- Tags: api, feature, integration, model
- URL: https://standarddb.com/slack/slack-launches-mcp-server-and-real-time-search-api-for-ai-agent-integration--fmpprfqv
- Original: https://docs.slack.dev/changelog/2026/02/17/slack-mcp

## New LLM-Focused Components

Slack announced the release of the **Slack MCP server** and **Real-time Search API**, two major additions designed to enhance how Large Language Models (LLMs) and AI agents interact with Slack workspace data.

### Slack MCP Server

The **Model Context Protocol (MCP) server** enables AI agents to interact with Slack content through tools purpose-built for LLM consumption. Unlike traditional APIs, the MCP server is specifically architected for language models, with robust descriptions and examples that return natural language responses. This allows AI agents to perform LLM-driven discovery, configuration, and execution within Slack.

### Real-time Search API

The **Data Access API has evolved into the Real-time Search API**, providing users with a secure search interface for accessing Slack data. This API allows third-party applications to retrieve relevant Slack data without storing customer information on external servers, reducing privacy and compliance concerns for enterprises.

### Granular Permission Scopes

The `assistant.search.context` API method now uses a set of granular scopes instead of a single `search:read` scope:

- **`search:read.public`** (required) — for public channel access
- **`search:read.private`** — for private channels (with user consent)
- **`search:read.im`** — for direct messages (with user consent)
- **`search:read.mpim`** — for multi-party direct messages (with user consent)

This enables more granular control over what data AI-enabled apps can access, improving security and user privacy while maintaining flexibility for developers.

### Getting Started

Developers can access the [Slack MCP server documentation](https://docs.slack.dev/ai/slack-mcp-server) and the [Real-time Search API guide](https://docs.slack.dev/apis/web-api/real-time-search-api) to begin integrating these capabilities into their applications.

---

### Plugins, Sandbox Access Controls, and Async Subagents
- Published: 2026-02-17T00:00:00+00:00
- Products: 
- Tags: release, feature, api, platform, integration
- URL: https://standarddb.com/cursor/cursor-2-5-launches-plugin-marketplace-and-async-subagents-for-multi-file--mtqkrgwd
- Original: https://cursor.com/changelog/2-5

## Plugins on the Cursor Marketplace

Cursor now supports plugins that package skills, subagents, MCP servers, hooks, and rules into single-install bundles. The new [Cursor Marketplace](https://cursor.com/marketplace) lets developers discover and install pre-built capabilities to extend the editor without manual configuration.

Initial plugin partners include **Amplitude, AWS, Figma, Linear, Stripe**, and more, covering workflows across design, databases, payments, analytics, and deployment. Install plugins directly from the editor using the `/add-plugin` command or browse available options at cursor.com/marketplace.

## Sandbox Network Access Controls

The sandbox environment now enforces granular network access controls alongside existing file and directory restrictions. Developers can define precisely which domains agents are allowed to reach:

- **User config only**: Restricted to domains specified in `sandbox.json`
- **User config with defaults**: Limited to custom allowlist plus Cursor's built-in defaults  
- **Allow all**: Unrestricted network access within the sandbox

Enterprise admins can enforce organization-wide network allowlists and denylists from the admin dashboard, ensuring consistent egress policies across all agent sandbox sessions.

## Async Subagents and Performance Improvements

Subagents previously ran synchronously, blocking the parent agent until completion. The new asynchronous execution model allows parent agents to continue working while subagents run in the background. Subagents can now spawn their own subagents, creating hierarchical task trees for tackling larger work like multi-file features, major refactors, and complex bug fixes.

Performance enhancements include reduced latency, better streaming feedback, and more responsive parallel execution across subagent operations.

---

### Assign issues to Copilot coding agent from Raycast
- Published: 2026-02-17T00:00:00+00:00
- Products: GitHub Copilot, GitHub Dependabot
- Tags: feature, integration, platform
- URL: https://standarddb.com/github/github-copilot-coding-agent-now-assignable-from-raycast-launcher--xbqpxgpf
- Original: https://github.blog/changelog/2026-02-17-assign-issues-to-copilot-coding-agent-from-raycast

## Raycast Integration for Copilot Coding Agent

GitHub has expanded the accessibility of its Copilot coding agent by integrating it with **Raycast**, a popular productivity launcher for macOS and Windows. Developers can now assign issues and manage coding tasks directly from Raycast using the GitHub Copilot extension, streamlining workflows without context-switching.

## How It Works

After installing the GitHub Copilot extension in Raycast, users gain access to three key commands:

- **Assign Issue to Copilot**: Select a repository and issue, then press `Cmd` + `Enter` (macOS) or `Ctrl` + `Enter` (Windows) to delegate the task. Copilot opens a pull request and works autonomously in the background.
- **Create Task**: Start new coding tasks without requiring an existing issue.
- **View Tasks**: Monitor the progress of active Copilot assignments in real-time.

Once assigned, Copilot operates asynchronously as a background agent, then requests your review when the work is complete.

## Availability and Requirements

The Copilot coding agent is available to **Copilot Pro**, **Copilot Pro+**, **Copilot Business**, and **Copilot Enterprise** subscribers. Copilot Business and Enterprise users must have the feature enabled by their administrator via the "Policies" page in their GitHub settings.

---

### Agents, Workers - Agents SDK v0.5.0: Protocol message control, retry utilities, data parts, and @cloudflare/ai-chat v0.1.0
- Published: 2026-02-17T00:00:00+00:00
- Products: Cloudflare Workers
- Tags: release, feature, sdk, api
- URL: https://standarddb.com/cloudflare/cloudflare-agents-sdk-v0-5-0-ships-retry-utilities-protocol-message-control-and--drlptkst
- Original: https://developers.cloudflare.com/changelog/2026-02-17-agents-sdk-v050

## Retry Utilities

Agents v0.5.0 introduces a new `this.retry()` method for retrying async operations with exponential backoff and jitter. Developers can pass an optional `shouldRetry` predicate to bail early on non-retryable errors, and configure `maxAttempts` per operation.

Retry options are available both at the method level and per-task for `queue()`, `schedule()`, `scheduleEvery()`, and `addMcpServer()`, with configurations persisted in SQLite. Class-level defaults can be set via `static options = { retry: { maxAttempts: 3 } }`, and invalid values throw immediately at enqueue/schedule time. Internal retries have also been added for workflow operations with Durable Object-aware error detection.

## Per-Connection Protocol Message Control

Agents automatically send JSON text frames (identity, state, MCP server lists) to every WebSocket connection. The new `shouldSendProtocolMessages()` method allows suppression of these messages per-connection for clients that cannot handle them — such as binary-only devices, MQTT clients, or lightweight embedded systems. The `isConnectionProtocolEnabled()` helper checks a connection's status at any time, and the flag persists across Durable Object hibernation.

## @cloudflare/ai-chat v0.1.0

The first stable release ships with a major internal refactor including a new `ResumableStream` class and simplified SSE parsing, with **zero breaking changes**. Key new features include:

- **Data parts** — Attach typed JSON blobs to messages alongside text with support for reconciliation, append, and transient parts
- **Tool approval persistence** — Approval UI survives page refresh and Durable Object hibernation via SQLite persistence
- **`maxPersistedMessages`** — Cap SQLite message storage with automatic oldest-message deletion
- **`body` option on `useAgentChat`** — Send custom data with every request
- **Incremental persistence** — Hash-based caching to skip redundant SQL writes
- **Row size guard** — Automatic two-pass compaction when messages approach the 2 MB SQLite limit
- **`autoContinueAfterToolResult` defaults to `true`** — Client-side tool results now automatically trigger server continuation

## Bug Fixes and Other Improvements

Notable fixes include resolved stream resumption race conditions, fixed `setMessages` functional updater emptying arrays, restored client tool schemas after Durable Object hibernation, resolved `InvalidPromptError` after tool approval, and fixed `reasoning-delta` data loss during stream resumption.

Synchronous queue and schedule getters (`getQueue()`, `dequeue()`, etc.) no longer return Promises despite performing only synchronous SQL operations, improving usability while maintaining backward compatibility. Additional improvements include a TypeScript depth counter preventing "excessively deep" type errors, POST SSE keepalive pings every 30 seconds, and widened peer dependency ranges to prevent cascading major bumps.

---

### Enterprise-wide credential management tools for incident response
- Published: 2026-02-17T00:00:00+00:00
- Products: GitHub
- Tags: feature, security, api
- URL: https://standarddb.com/github/github-enterprise-adds-credential-management-tools-for-incident-response--lvwcqbgv
- Original: https://github.blog/changelog/2026-02-17-enterprise-wide-credential-management-tools-for-incident-response

## New Security Incident Response Capabilities

GitHub Enterprise Cloud has introduced enterprise-wide credential management tools designed to help organizations respond quickly and decisively to compromised credentials and high-impact security incidents.

### Key Features

Enterprise owners (and delegated administrators with the new `Manage enterprise credentials` permission) can now:

- **Review credential counts**: View authorized credentials via SSO across one or more organizations in your enterprise
- **Temporarily block SSO**: Restrict SSO access to enterprise owners only, reducing blast radius during investigation
- **Revoke SSO authorizations**: Invalidate personal access tokens, SSH keys, and OAuth tokens across the enterprise
- **Delete tokens and SSH keys**: Remove credentials entirely across the enterprise (EMU accounts only)

### Availability and Scope

These capabilities are available for:
- Enterprise Managed Users (EMU)
- Enterprises with personal accounts that have enabled single sign-on (SSO) at the enterprise or organization level

Enterprise owners can delegate credential management responsibilities to trusted administrators by assigning the new fine-grained `Manage enterprise credentials` permission.

### Important Considerations

**Use during major incidents only.** These actions can break automations and disrupt developer workflows, so GitHub recommends using them only when responding to active security threats. For routine credential hygiene, enterprise administrators should use GitHub's [maximum token lifetime policies](https://docs.github.com/enterprise-cloud@latest/admin/enforcing-policies/enforcing-policies-for-your-enterprise/enforcing-policies-for-personal-access-tokens-in-your-enterprise) instead.

All credential management actions are logged in the audit log, providing enterprise owners with detailed context for forensic investigation and compliance purposes.

---

### Access - Policies for bookmark applications
- Published: 2026-02-17T00:00:00+00:00
- Products: Cloudflare
- Tags: feature, api, platform
- URL: https://standarddb.com/cloudflare/cloudflare-access-now-supports-policies-for-bookmark-applications--rvtrgdrp
- Original: https://developers.cloudflare.com/changelog/post/2026-02-17-policies-for-bookmarks

## Access Policies Now Available for Bookmarks

Cloudflare has extended its Access policy framework to bookmark applications, enabling fine-grained control over bookmark visibility in the App Launcher. This enhancement allows organizations to tailor application access based on user identity, device posture, group membership, and other policy criteria.

### What Changed

Bookmark applications can now have Access policies assigned to them, giving administrators the ability to:

- **Control visibility per user** — Users see only the bookmarks they're authorized to access, reducing confusion and improving security by preventing accidental exposure to sensitive resources.
- **Restrict sensitive resources** — Bookmarks to internal tools, partner portals, or restricted services can be limited to specific groups, identity providers, or devices meeting particular posture requirements.

### Supported Configurations

Bookmarks support all standard Access policy configurations except:
- Purpose justification
- Temporary authentication
- Application isolation

Backward compatibility is maintained: bookmarks without an assigned policy remain visible to all users in the organization.

### Getting Started

To add policies to bookmarks, navigate to your Access policies configuration and apply the same rules you'd use for other protected applications. Refer to the [Add bookmarks documentation](https://developers.cloudflare.com/cloudflare-one/access-controls/applications/bookmarks/) for detailed instructions.

---

### Containers - Docker-in-Docker support added to Containers and Sandboxes
- Published: 2026-02-17T00:00:00+00:00
- Products: Cloudflare
- Tags: feature, platform, sdk
- URL: https://standarddb.com/cloudflare/cloudflare-adds-docker-in-docker-support-to-containers-and-sandboxes--ypjbshsp
- Original: https://developers.cloudflare.com/changelog/2026-02-17-docker-in-docker

## Docker-in-Docker Now Available

Cloudflare has announced Docker-in-Docker support for both Containers and Sandboxes platforms. This feature enables developers to run Docker environments within containerized sandboxes, opening up new workflows for application development and testing.

## Use Cases

This capability supports several important development scenarios:

- **Containerized Application Development**: Develop and test containerized applications directly within your Sandbox environment
- **Isolated Test Environments**: Run isolated test suites against Docker images without external dependencies
- **CI/CD Integration**: Build and compile container images as part of continuous integration and deployment workflows
- **Runtime Flexibility**: Deploy arbitrary container images supplied at runtime, enabling dynamic workload management

## Implementation Details

For developers using the Sandbox SDK, Cloudflare has published a dedicated [Docker-in-Docker guide](https://developers.cloudflare.com/sandbox/guides/docker-in-docker/) with implementation instructions. Developers working with Containers directly can consult the [Containers FAQ](https://developers.cloudflare.com/containers/faq/#can-i-run-docker-inside-a-container-docker-in-docker) for configuration details.

This feature is particularly valuable for teams running agents or end users who need complete, isolated development environments with full containerization capabilities.

---

### Claude Sonnet 4.6 is now generally available in GitHub Copilot
- Published: 2026-02-17T00:00:00+00:00
- Products: GitHub Copilot
- Tags: release, feature, model, api
- URL: https://standarddb.com/github/claude-sonnet-4-6-now-available-in-github-copilot-for-all-subscription-tiers--swgqhqmz
- Original: https://github.blog/changelog/2026-02-17-claude-sonnet-4-6-is-now-generally-available-in-github-copilot

## Claude Sonnet 4.6 General Availability

Anthropic's Claude Sonnet 4.6, an agentic coding model, is now rolling out in GitHub Copilot. Early testing shows the model excels at agentic coding tasks and demonstrates particular strength in search operations.

## Where to Find It

The model is available to all Copilot Pro, Pro+, Business, and Enterprise subscribers and can be selected via the model picker in:
- **Visual Studio Code** (chat, ask, edit, agent modes)
- **Visual Studio** (agent, ask modes)
- **github.com**
- **GitHub Mobile** (iOS and Android)
- **GitHub Copilot CLI**
- **GitHub Copilot Coding Agent**

The rollout is gradual, so you may need to check back if the model isn't immediately available in your account.

## Important Notes for Admins

Copilot Enterprise and Business plan administrators must explicitly enable the Claude Sonnet 4.6 policy in Copilot settings to make the model available to their teams.

## Pricing

While Claude Sonnet 4.6 is launching with a **1x premium request multiplier**, GitHub notes that pricing is tentative and subject to change. This means request costs are currently at standard rates, but this may be adjusted as the service matures.

For a complete list of available models in GitHub Copilot, refer to the [official documentation on supported models](https://docs.github.com/copilot/reference/ai-models/supported-models).

---

### Custom properties and rule insights improvements
- Published: 2026-02-17T00:00:00+00:00
- Products: GitHub
- Tags: feature, platform
- URL: https://standarddb.com/github/github-enhances-custom-properties-and-rule-insights-with-mandatory-value--vxgmyqpv
- Original: https://github.blog/changelog/2026-02-17-custom-properties-and-rule-insights-improvements

## Custom Properties Now Enforce Explicit Values

Organization and enterprise administrators can now require repository creators to explicitly select values for custom properties, rather than relying on default values. This ensures repositories meet policy requirements from the moment they're created, improving adoption of rulesets and data accuracy across your organization.

Previously, the asterisk (`*`) on the repository creation page indicated a required property that could accept a default value. With this update, the asterisk now clearly indicates properties requiring explicit selection, giving administrators tighter control over repository configuration and ensuring no policies are inadvertently bypassed through defaults.

## Rule Insights Get Dynamic Filtering with SHA Support

The rule insights experience has been improved with a new dynamic filter bar replacing the previous dropdown-based filters. The key enhancement is support for filtering by commit SHA on the repository-level rule insights page, making it significantly easier to understand how specific commits interacted with your rulesets.

This more flexible filtering approach provides a consistent experience and better visibility into rule activity, helping teams debug issues and understand ruleset behavior across their repositories.

## What You Need to Do

- **For custom properties**: Review your organization's custom property configurations to determine which properties should require explicit selection versus defaults
- **For rule insights**: Leverage the new SHA filtering to troubleshoot specific commits and verify ruleset compliance

---

### Sonnet 4.6, Now Live in Harvey
- Published: 2026-02-17T00:00:00+00:00
- Products: 
- Tags: release, model, feature
- URL: https://standarddb.com/harvey/harvey-deploys-claude-sonnet-4-6-scores-87-6-on-legal-tasks--fcgxgrnq
- Original: https://www.harvey.ai/blog/sonnet-46-now-live-in-harvey

## Claude Sonnet 4.6 Now Available in Harvey

Harvey has made Claude Sonnet 4.6—Anthropic's latest model in the Sonnet family—available to eligible clients within its legal AI platform. Sonnet 4.6 combines strong reasoning capabilities with a faster, more efficient architecture designed for practical problem-solving and instruction-following.

### Performance Metrics

On Harvey's proprietary **BigLaw Bench evaluation suite**, Sonnet 4.6 achieved an **87.6% overall score**, with 35% of tasks receiving perfect scores. Performance was balanced across practice areas, with equivalent average scores for both litigation and transactional legal work.

### Key Strengths

Harvey's Applied Legal Research team identified several areas where Sonnet 4.6 excels:

- **Numerical precision**: Delivers more accurate specific figures (e.g., dollar amounts) when explicitly instructed compared to prior Claude models
- **Ideation tasks**: Shows clear advantages in thoroughness for trial exhibit preparation and open-ended litigation strategy questions
- **Code analysis**: Excels at root cause analysis in Claude Code, correctly tracing bugs through code history and integrating parallel tool calls to solve problems that even Opus 4.6 missed

### Important Usage Pattern

Evaluators noted an important nuance: Sonnet 4.6 performs best when given explicit direction (retrieve exact figures, follow a defined rubric, produce structured comparisons). When left to determine scope independently, it tends to remain at higher summary levels, sometimes missing fine-grained details—a gap most visible on multi-document synthesis tasks.

### Rollout

Harvey is rolling out Sonnet 4.6 in the model selector to eligible clients, with broader availability expected to follow.

---

### API access to billing usage reports in public preview
- Published: 2026-02-17T00:00:00+00:00
- Products: GitHub
- Tags: api, feature, release
- URL: https://standarddb.com/github/github-releases-billing-usage-reports-rest-api-in-public-preview--wcjlzhnq
- Original: https://github.blog/changelog/2026-02-17-api-access-to-billing-usage-reports-in-public-preview

## Overview

GitHub Enterprise administrators now have programmatic access to billing usage reports through a new REST API currently in public preview. This capability allows teams to automate the retrieval and processing of billing data without manual downloads from the GitHub UI.

## What's New

The new **Usage Reports API** provides REST endpoints for:
- Requesting billing usage reports programmatically
- Downloading reports in CSV format
- Accessing all existing billing report types available in the UI (such as license usage, commit, and member activity reports)
- Automating billing data workflows for enterprise accounts

## How to Get Started

- Review the complete API documentation in the [REST API docs for usage reports](https://docs.github.com/en/enterprise-cloud@latest/rest/billing/usage-reports)
- Example scripts and implementation guides are available in the [GitHub Community discussions](https://github.com/orgs/community/discussions/186162)
- Provide feedback on the public preview to help shape the final release

## What Developers Need to Know

This API is currently in **public preview**, meaning the interface may change before general availability. Enterprise Cloud administrators should test integration workflows early to ensure compatibility with their billing systems and automation tools.

---

### Codex CLI Release: 0.103.0
- Published: 2026-02-17T00:00:00+00:00
- Products: OpenAI API
- Tags: release, feature, api, sdk, platform
- URL: https://standarddb.com/openai/codex-cli-0-106-0-ships-with-native-install-script-experimental-realtime-thread--zntfbygy
- Original: https://developers.openai.com/codex/changelog

## Key Updates

**Installation and Platform Support**
- Added a direct install script for macOS and Linux, distributed as a GitHub release asset using the existing platform payload (including `codex` and `rg` tools)
- Installation via npm remains available: `npm install -g @openai/codex@0.106.0`

**App-Server v2 Enhancements**
The app-server v2 thread API now includes:
- Experimental thread-scoped realtime endpoints and notifications for live collaboration
- New `thread/unsubscribe` flow allowing developers to unload threads without archiving them
- These additions enable more flexible real-time interaction patterns within threaded conversations

**JavaScript REPL Promotion**
- JavaScript REPL functionality promoted from `/unstable` to `/experimental` status
- Added startup compatibility checks with user-visible warnings to help developers understand system requirements
- Lowered validated minimum Node.js version requirement to `22.22.0` for broader compatibility

**Collaboration and Model Access**
- Enabled `request_user_input` capability in Default collaboration mode (previously limited to Plan mode)
- Made `5.3-codex` model visible in CLI model list for API users, improving model discovery
- Improved memory behavior through diff-based forgetting and usage-aware memory selection for more efficient context management

## Getting Started

Update to version 0.106.0 using npm or by using the new native install script for macOS/Linux systems. Review the updated compatibility requirements for JavaScript REPL if you use that feature.

---

### Access - Policies for bookmark applications
- Published: 2026-02-17T00:00:00+00:00
- Products: Cloudflare
- Tags: feature, api
- URL: https://standarddb.com/cloudflare/cloudflare-access-now-supports-policies-for-bookmark-applications--ygjvxbyt
- Original: https://developers.cloudflare.com/changelog/2026-02-17-policies-for-bookmarks

## Overview

Cloudflare Access now supports policies for bookmark applications, bringing fine-grained access control to bookmarked resources in the App Launcher. Previously, bookmark applications were visible to all users in your organization, but now administrators can assign the same Access policies used for other applications to control visibility.

## Key Capabilities

With policy support for bookmarks, you can now:

- **Tailor the App Launcher to each user** — Users see only the applications they have access to, reducing clutter and preventing accidental clicks on irrelevant resources
- **Restrict visibility of sensitive bookmarks** — Limit bookmark visibility to internal tools or partner resources based on group membership, identity provider, or device posture
- **Maintain backwards compatibility** — If no policy is assigned to a bookmark, it remains visible to all users

## Supported Policy Features

Bookmarks support all Access policy configurations with the exception of:
- Purpose justification
- Temporary authentication
- Application isolation

## Getting Started

For detailed information on how to add bookmarks with policies, refer to the [Add bookmarks documentation](https://developers.cloudflare.com/cloudflare-one/access-controls/applications/bookmarks/).


---

### Agents, Workers - Agents SDK v0.5.0: Protocol message control, retry utilities, data parts, and @cloudflare/ai-chat v0.1.0
- Published: 2026-02-17T00:00:00+00:00
- Products: Cloudflare Workers
- Tags: release, feature, sdk, api
- URL: https://standarddb.com/cloudflare/cloudflare-agents-sdk-v0-5-0-adds-retry-utilities-protocol-message-control-and--dfjpbqyh
- Original: https://developers.cloudflare.com/changelog/post/2026-02-17-agents-sdk-v050

## Retry Utilities

A new `this.retry()` method allows you to retry any async operation with exponential backoff and jitter. You can optionally pass a `shouldRetry` predicate to bail early on non-retryable errors. Retry options are also available per-task on `queue()`, `schedule()`, `scheduleEvery()`, and `addMcpServer()`, with configurations persisted in SQLite alongside tasks. Class-level retry defaults can be set via `static options`, and all options are validated eagerly at enqueue/schedule time. Internal retries have been added for workflow operations with Durable Object-aware error detection.

## Per-Connection Protocol Message Control

Agents automatically send JSON text frames to every WebSocket connection. You can now suppress these protocol messages per-connection via the `shouldSendProtocolMessages()` hook—useful for binary-only devices, MQTT clients, or lightweight embedded systems. Connections with protocol messages disabled still fully participate in RPC and regular messaging, and you can check a connection's status using `isConnectionProtocolEnabled()`. The flag persists across Durable Object hibernation.

## @cloudflare/ai-chat v0.1.0

The first stable release of `@cloudflare/ai-chat` includes a major internal refactor with zero breaking changes. Existing code using `AIChatAgent` and `useAgentChat` works without modification.

**New capabilities:**
- **Data parts** — Attach typed JSON blobs (`data-*`) to messages alongside text, with support for reconciliation, append, and transient parts
- **Tool approval persistence** — The approval UI now survives page refresh and Durable Object hibernation by persisting messages to SQLite
- **`maxPersistedMessages`** — Cap SQLite storage with automatic oldest-message deletion
- **`body` option on `useAgentChat`** — Send custom data with every request (static or dynamic)
- **`autoContinueAfterToolResult` defaults to `true`** — Client-side tool results and approvals now automatically trigger server continuation

## Additional Improvements

- **Synchronous queue/schedule getters** — Methods like `getQueue()`, `getSchedule()`, and `dequeue()` are now synchronous and return values directly instead of Promises. Existing code using `await` remains compatible.
- **TypeScript depth fix** — A depth counter on `CanSerialize` types prevents "excessively deep" instantiation errors with nested structures like `CoreMessage[]`
- **POST SSE keepalive** — The POST SSE handler now sends `event: ping` every 30 seconds to prevent connection drops during long-running tool calls
- **Widened peer dependencies** — Peer dependency ranges have been expanded to prevent cascading major bumps, with `@cloudflare/ai-chat` and `@cloudflare/codemode` marked as optional

Notable bug fixes address stream resumption race conditions, `setMessages` functional updater issues, client tool schema loss after hibernation, and message metadata propagation on broadcast/resume paths.

---

### MCP Registry and more improvements in Copilot in Eclipse
- Published: 2026-02-17T00:00:00+00:00
- Products: GitHub Copilot
- Tags: feature, release, platform, integration
- URL: https://standarddb.com/github/github-copilot-in-eclipse-gains-mcp-registry-todo-management-and-ui-improvements--sblxwfjm
- Original: https://github.blog/changelog/2026-02-17-mcp-registry-and-more-improvements-in-copilot-in-eclipse

## New Features for Eclipse Integration

GitHub Copilot in Eclipse now includes an **MCP Registry**—a centralized hub for discovering and installing Model Context Protocol servers without manual configuration. Developers can browse available servers, review their capabilities, and add them to their workspace with just a few clicks.

The update also introduces a **Manage Todo List tool** that helps organize complex development tasks. As Copilot works through multi-step requests, it creates and updates a structured todo list in real-time, providing clear visibility into progress and remaining work.

## Enhanced User Experience

The chat interface has received significant polish:

- **Font size control**: Adjust text size using `⌘ + =` / `⌘ + -` (macOS) or `Ctrl + =` / `Ctrl + -` (Windows/Linux)
- **Dark theme refresh**: Improved contrast and readability for extended coding sessions
- **Undo/Redo support**: Correct chat input mistakes seamlessly
- **Native toolbar integration**: Chat controls now integrate naturally with Eclipse's main toolbar instead of floating on the chat view

## Smarter Context and Customization

**Editor Selection Context** automatically includes code selections in chat conversations, making interactions more focused and relevant without additional manual steps.

New preference options give developers more control:

- **Agent max requests**: Limit how many requests an agent can execute before asking to continue, providing boundaries for complex tasks
- **Commit instructions**: Customize Copilot's commit message generation to match team conventions and style guides

To ensure the new toolbar buttons display correctly, users may need to delete the `workbench.xmi` file at `/.metadata/.plugins/org.eclipse.e4.workbench/` if buttons don't appear after updating.

---

### Containers - Docker-in-Docker support added to Containers and Sandboxes
- Published: 2026-02-17T00:00:00+00:00
- Products: Cloudflare Workers, Cloudflare
- Tags: feature, platform, sdk
- URL: https://standarddb.com/cloudflare/cloudflare-containers-and-sandboxes-gain-docker-in-docker-support--xzbwvpzj
- Original: https://developers.cloudflare.com/changelog/post/2026-02-17-docker-in-docker

## Docker-in-Docker Now Available

Cloudflare has added Docker-in-Docker (DinD) support to both its Containers and Sandboxes services. This capability allows developers to run Docker inside containerized or sandboxed environments, enabling more complex application development and deployment scenarios.

## Use Cases

With this new feature, you can now:

- **Develop containerized applications** within your Sandbox environment
- **Run isolated test environments** for container images
- **Build container images** as part of CI/CD workflows
- **Deploy arbitrary images** supplied at runtime within a container

This is particularly valuable for teams using agents or end users who need complete, isolated development environments with full Docker capabilities.

## Getting Started

**For Sandbox SDK users:** Refer to the [Docker-in-Docker guide](https://developers.cloudflare.com/sandbox/guides/docker-in-docker/) in the Cloudflare documentation for integration instructions.

**For Containers users:** Check the [Containers FAQ](https://developers.cloudflare.com/containers/faq/#can-i-run-docker-inside-a-container-docker-in-docker) for implementation details and best practices.

This enhancement expands the flexibility of Cloudflare's edge computing platform, allowing developers to run more sophisticated workloads on the edge with full containerization support.

---

### Access - Streamlined clientless access for private applications
- Published: 2026-02-17T00:00:00+00:00
- Products: Cloudflare
- Tags: feature, platform
- URL: https://standarddb.com/cloudflare/cloudflare-access-adds-streamlined-clientless-access-for-private-applications--wxdqlcsz
- Original: https://developers.cloudflare.com/changelog/post/2026-02-17-clientless-access-for-private-apps

## Simplified Clientless Access Management

Cloudflare has streamlined its approach to providing clientless access for private applications within its Access service. The new **Allow clientless access** setting eliminates the need for administrators to create separate bookmark applications and manage complex URL prefixes.

## How It Works

**Before this update**, organizations had to:
- Create a separate bookmark application
- Manually construct prefixed Clientless Web Isolation URLs (e.g., `https://<your-teamname>.cloudflareaccess.com/browser/https://10.0.0.1/`)
- Expose this bookmark to all users in the App Launcher, even those without underlying application access

**Now**, administrators can:
- Enable **Allow clientless access** directly in private self-hosted application settings
- Have users who pass Access policies automatically see the application tile in their App Launcher
- Maintain consistent access control — only users with both Access application permissions AND remote browser permissions can access the link

## User Experience Improvements

Users now benefit from a more intuitive workflow: applications appear in their App Launcher with proper access controls applied, and the clientless access mechanism is transparent. The integration ensures that access policies are consistently enforced across both traditional and clientless access methods.

---

### Delegate tasks to Copilot coding agent from Visual Studio
- Published: 2026-02-17T00:00:00+00:00
- Products: GitHub Copilot, GitHub
- Tags: release, feature, integration
- URL: https://standarddb.com/github/github-copilot-coding-agent-now-available-in-visual-studio-with-new-task--njwmgnrh
- Original: https://github.blog/changelog/2026-02-17-delegate-tasks-to-copilot-coding-agent-from-visual-studio

## Visual Studio Integration for Copilot Coding Agent

GitHub's Copilot coding agent—an asynchronous, autonomous background agent that creates draft pull requests and works in the cloud—is now accessible directly from Visual Studio. This expansion brings the task delegation workflow to one of the most popular development environments.

### How It Works

To delegate a task, developers simply:
- Enter a prompt in Copilot Chat within Visual Studio
- Click the new **Send to Copilot Coding Agent** button next to the Send button
- Confirm the action when prompted
- Copilot opens a pull request and begins working in the background
- The agent requests a review once the work is complete

### Requirements and Availability

**Subscription Tiers:** Available to Copilot Pro, Copilot Pro+, Copilot Business, and Copilot Enterprise subscribers.

**Software Requirements:**
- Visual Studio 2026 with at least December Update 18.1.0
- The **Enable Copilot Coding agent (preview)** setting must be enabled in Visual Studio

**Enterprise Configuration:** Copilot Business and Copilot Enterprise administrators must enable the Copilot coding agent from the "Policies" page before team members can use the feature.

### Getting Started

To learn more about implementing this workflow, refer to the GitHub documentation on ["Asking Copilot to create a pull request from Copilot Chat in Visual Studio 2026"](https://docs.github.com/copilot/how-tos/use-copilot-agents/coding-agent/create-a-pr#asking-copilot-to-create-a-pull-request-from-copilot-chat-in-visual-studio-2026).

---

### Qwen 3.5 Plus is on AI Gateway
- Published: 2026-02-16T13:00:00+00:00
- Products: Vercel AI SDK, Vercel
- Tags: feature, integration, model, api
- URL: https://standarddb.com/vercel/qwen-3-5-plus-now-available-on-vercel-s-ai-gateway-with-1m-context-window--fgxvtkmr
- Original: https://vercel.com/changelog/qwen-3-5-plus-is-on-ai-gateway

## Qwen 3.5 Plus Now Available

Alibaba's Qwen 3.5 Plus model is now accessible through Vercel's AI Gateway, bringing new capabilities for developers building AI-powered applications. The model features:

- **1M context window** for handling large amounts of input data
- **Adaptive tool use** for orchestrating complex workflows
- **Multimodal capabilities** for processing both text and visual inputs
- **Agentic workflows** suitable for autonomous agent applications

## Key Capabilities and Use Cases

Qwen 3.5 Plus is optimized for several developer-focused scenarios:

- Web development and frontend task automation
- Converting design mockups into production-ready React components
- Scientific problem solving and visual reasoning
- Multi-step workflows requiring tool orchestration

The model delivers stronger performance compared to Qwen 3 VL in scientific problem solving and visual reasoning tasks.

## Integration with AI Gateway

To use Qwen 3.5 Plus, developers can access it through Vercel's AI SDK by setting the model parameter to `alibaba/qwen3.5-plus`. This provides access to AI Gateway's unified interface, which includes built-in observability, usage tracking, cost management, intelligent provider routing, and automatic retry logic.

Developers can explore the model in Vercel's model playground and compare performance across other available models on the AI Gateway leaderboard.

---

### GLM-5 and Minimax M2.5 are now available in Windsurf
- Published: 2026-02-16T12:00:00+00:00
- Products: 
- Tags: 
- URL: https://standarddb.com/windsurf/glm-5-and-minimax-m2-5-are-now-available-in-windsurf--plyzgdwr
- Original: https://windsurf.com/blog/glm-5-minimax-m2.5



---

### Qwen3.5 is here!
- Published: 2026-02-16T00:00:00+00:00
- Products: 
- Tags: release, model, feature, open-source
- URL: https://standarddb.com/unsloth/alibaba-releases-qwen3-5-model-family-397b-variant-matches-gemini-3-pro-and--qgpgdxwy
- Original: https://unsloth.ai/docs/models/qwen3.5

## Qwen3.5 Model Family Launch

Alibaba has released **Qwen3.5**, a new multimodal model family with four size variants:

- **Qwen3.5-27B**: Compact option for resource-constrained deployments
- **Qwen3.5-35B-A3B**: Balanced model with 3B active parameters
- **Qwen3.5-122B-A10B**: Mid-tier with 10B active parameters  
- **Qwen3.5-397B-A17B**: Flagship 397B model with only 17B active parameters (mixture-of-experts)

The models support **256K token context window** (extendable to 1M), operate in 201 languages, and include hybrid **thinking and non-thinking modes** for flexible reasoning capabilities.

## Performance & Capabilities

The flagship **Qwen3.5-397B-A17B** delivers performance comparable to **Gemini 3 Pro**, **Claude Opus 4.5**, and **GPT-5.2**. Strengths include:

- **Coding**: Advanced programming task support
- **Vision**: Multimodal understanding capabilities
- **Agents**: Autonomous task execution with tool calling
- **Long-context**: Efficient processing of extended documents

## Deployment & Quantization

Unsloth provides **Dynamic 2.0 quantized versions** optimized for local deployment:

- **4-bit MXFP4** (~214GB): Runs on 256GB RAM/Mac M3 Ultra with 25+ tokens/s
- **3-bit quantization** (~192GB): Fits on 192GB RAM devices
- **Full precision** (~807GB): Requires 512GB+ memory

All quantizations use dynamic casting to preserve critical layers at 8-16 bit precision. Complete models are available on Hugging Face, with tutorials for **llama.cpp**, **llama-server**, and **OpenAI API compatibility**.

## Developer Integration

The release includes comprehensive deployment guides for local inference using llama.cpp with configurable parameters for thinking vs. non-thinking modes. Tool calling support enables building autonomous agents with custom functions. Maximum recommended output is 32,768 tokens per query.

---

### Cloudflare Fundamentals - Content encoding support for Markdown for Agents and other improvements
- Published: 2026-02-16T00:00:00+00:00
- Products: Cloudflare
- Tags: feature, api, platform
- URL: https://standarddb.com/cloudflare/cloudflare-boosts-markdown-for-agents-with-content-encoding-support-doubles--svhnptvc
- Original: https://developers.cloudflare.com/changelog/post/2026-02-16-markdown-for-agents-improvements

## Markdown for Agents Enhancement

Cloudflare has released improvements to its **Markdown for Agents** feature, enabling AI systems to request web pages in Markdown format instead of HTML. When enabled on your zone, the feature allows AI crawlers to express a preference for `text/markdown` in their requests, and Cloudflare's network automatically converts HTML to Markdown on the fly.

## Key Improvements

This release delivers three important enhancements:

- **Origin response limit increased**: The maximum origin response size has been doubled from 1 MB to 2 MB (2,097,152 bytes), enabling conversion of larger pages
- **Removed content-length requirement**: The origin no longer needs to send the `content-length` header, improving compatibility with more servers
- **Content encoding support**: The feature now handles content-encoded (gzip, brotli, etc.) responses from the origin, expanding coverage across diverse backend configurations

## Getting Started

To enable this feature, navigate to the **AI Crawl Control** section in your [Cloudflare dashboard](https://dash.cloudflare.com/?to=/:account/:zone/ai) and enable **Markdown for Agents**. This allows AI systems like search engines, LLMs, and other intelligent bots to receive cleaner, more structured content that's easier to parse and process.

For detailed implementation guidance and configuration options, refer to the [Markdown for Agents developer documentation](https://developers.cloudflare.com/fundamentals/reference/markdown-for-agents/).

---

### WAF - WAF Release - 2026-02-16
- Published: 2026-02-16T00:00:00+00:00
- Products: Cloudflare
- Tags: security, bugfix
- URL: https://standarddb.com/cloudflare/cloudflare-waf-adds-protections-for-zimbra-lfi-and-vite-path-traversal-cves--tshwfwns
- Original: https://developers.cloudflare.com/changelog/post/2026-02-16-waf-release

## New CVE Detections

Cloudflare's WAF has been updated with new detection and blocking rules for two recently disclosed vulnerabilities affecting widely-used software.

### CVE-2025-68645: Zimbra Local File Inclusion
The first vulnerability affects Zimbra Collaboration Suite (ZCS) versions 10.0 and 10.1. This Local File Inclusion (LFI) vulnerability allows unauthenticated attackers to craft malicious requests to the `/h/rest` endpoint, which can improperly influence internal request dispatching and lead to arbitrary file inclusion from the WebRoot directory. This poses a significant risk to organizations running affected versions.

### CVE-2025-31125: Vite Path Traversal
The second CVE targets Vite, the popular JavaScript frontend build tool. When Vite's development server is exposed to the network, attackers can exploit the `?inline&import` query parameters to read non-allowed files, potentially exposing sensitive information and configuration data.

## WAF Rule Updates

Two new rules have been added to the Cloudflare Managed Ruleset:

- **Zimbra LFI Detection** (Rule ID: ...833761f7): Detects and blocks attempts to exploit the Zimbra vulnerability. Previous action was logging only; now set to block.
- **Vite Path Traversal Detection** (Rule ID: ...950ed8c8): Detects and blocks attempts to exploit the Vite vulnerability. Previous action was logging only; now set to block.

Both rules are now active and blocking by default, providing immediate protection for customers without requiring additional configuration.

---

### February 16, 2026
- Published: 2026-02-16T00:00:00+00:00
- Products: 
- Tags: release, feature, api, sdk
- URL: https://standarddb.com/elevenlabs/elevenlabs-expands-elevenagents-with-versioning-rag-tools-and-content-guardrails--sxxydqwt
- Original: https://elevenlabs.io/docs/changelog/2026/2/16

## Agent Versioning and Configuration

ElevenLabs introduced agent versioning capabilities, allowing developers to fetch and update specific agent configurations using new `version_id` and `branch_id` query parameters. The [Get agent](https://elevenlabs.io/docs/api-reference/agents/get) endpoint now supports these parameters for version-specific retrieval, while the [Update agent](https://elevenlabs.io/docs/api-reference/agents/update) endpoint accepts `branch_id` for branch-scoped modifications. This enables better version control and branch-based development workflows for agent configurations.

## RAG and Tool Expansion

A new `search_documentation` built-in system tool provides retrieval-augmented generation (RAG) capabilities with multi-source document retrieval. Developers can configure multiple sources via `MultiSourceConfigJson`, `SourceConfigJson`, and `SourceRetrievalConfig` schemas, with customizable result merging strategies. Additionally, **MCP (Model Context Protocol) tool support** is now available as a new tool type alongside existing webhook, client, and system tools, expanding agent extensibility options.

## Content Moderation and Safety

Content guardrails have been added to agent configurations with configurable thresholds across seven categories: sexual content, violence, harassment, self-harm, profanity, religion/politics, and medical/legal content. The new `ContentGuardrail` and `ContentConfig` schemas allow fine-grained control over content moderation behavior. An `expressive_mode` boolean field enables agents to leverage the new v3 conversational model for more natural interactions.

## Enhanced Conversation and Testing Features

A new endpoint (`GET /v1/convai/users`) lists users who have had conversations with agents, supporting pagination and filtering by agent, time range, and other criteria. Phone number transfers now support `post_dial_digits` for sending DTMF tones after connection. Agent testing was restructured into three distinct types: `llm` (response evaluation), `tool` (tool-call verification), and `simulation` (full conversation simulation) with dedicated schemas for each type.

## SDK and Pronunciation Dictionary Updates

Python and JavaScript SDKs updated to v2.36.0 with support for all new features. The JavaScript SDK added overloaded `convert` signatures to `speechToText.convert()` for improved type inference. Widget packages (@elevenlabs/convai-widget-core and @elevenlabs/convai-widget-embed) now display agent tool usage status and long-running tool call badges. The [Get pronunciation dictionary](https://elevenlabs.io/docs/api-reference/pronunciation-dictionaries/get) endpoint now returns full rule details in a `rules` array, and folder deletion support was added to the knowledge base API with recursive deletion capability via the `force` parameter.

---

### SGLang-Diffusion: Advanced Optimizations for Production-Ready Video Generation
- Published: 2026-02-16T00:00:00+00:00
- Products: 
- Tags: feature, performance, sdk, open-source
- URL: https://standarddb.com/lmsys/sglang-diffusion-v2-adds-token-level-sharding-and-parallel-vae-for-video--sjjycvcx
- Original: https://lmsys.org/blog/2026-02-16-sglang-diffusion-advanced-optimizations

## Token-Level Sequence Sharding

SGLang-Diffusion improves sequence parallel sharding by flattening the temporal, height, and width dimensions (T × H × W) into a single sequence dimension before sharding across GPUs. This replaces the previous frame-level approach, which required padding frames and introduced significant computational overhead. For typical video resolutions, the new token-level strategy eliminates padding entirely while reducing all-to-all communication volume by up to 12.5%.

## Distributed VAE and Parallel Folding

The framework now supports **Parallel VAE** with height-wise sharding for high-resolution video encoding/decoding, using `halo_exchange` for boundary pixel sharing and `all_gather` for attention operations. Additionally, **Parallel Folding** decouples Text Encoder and DiT parallelism strategies, allowing the Text Encoder to reuse the DiT's sequence parallel group as tensor parallelism, improving memory efficiency without sacrificing throughput.

## Cache-DiT Multi-Request Stability

Critical bugs in Cache-DiT serving have been fixed:
- Each transformer in the dual-transformer architecture (`transformer` and `transformer_2`) now independently manages cache contexts using their own step counts
- Cache contexts are refreshed for each new request, preventing buffer contamination and shape mismatch crashes when handling variable video shapes

## I/O and Kernel Optimizations

Video save operations now run directly on GPU workers, eliminating serialization/deserialization overhead and tensor copies. Additionally, custom JIT kernels for WanVideo LayerNorm variants (including `LayerNormScaleShift` and `ScaleResidualLayerNormScaleShift`) reduce GPU bubbles by fusing elementwise operations with normalization reductions.

---

### Cloudflare Fundamentals - Content encoding support for Markdown for Agents and other improvements
- Published: 2026-02-16T00:00:00+00:00
- Products: Cloudflare
- Tags: feature, api, platform
- URL: https://standarddb.com/cloudflare/cloudflare-markdown-for-agents-now-supports-content-encoding-raises-1mb-limit--wfntqgzd
- Original: https://developers.cloudflare.com/changelog/2026-02-16-markdown-for-agents-improvements

## Markdown for Agents Enhancements

Cloudflare's **Markdown for Agents** feature enables AI systems to request website content in markdown format instead of HTML, making it easier for large language models to parse and process web pages. Today's release includes significant improvements to the underlying infrastructure that powers this feature.

### Key Improvements

- **Increased response limit**: The maximum origin response size has been doubled from 1 MB to 2 MB (2,097,152 bytes), allowing larger pages to be converted to markdown
- **Flexible headers**: No longer requires the origin to send the `content-length` header, improving compatibility with more origin servers
- **Content encoding support**: Now handles content-encoded responses (such as gzip-compressed content) from origins, automatically decompressing before conversion

### Getting Started

If you haven't enabled this feature, navigate to the **AI Crawl Control** section in your Cloudflare dashboard and enable **Markdown for Agents**. Once enabled, AI systems requesting your pages can signal their preference for `text/markdown` in HTTP Accept headers, and Cloudflare's network will automatically convert HTML to markdown on the fly.

For implementation details and full documentation, refer to the [Markdown for Agents developer documentation](https://developers.cloudflare.com/fundamentals/reference/markdown-for-agents/).

---

### WAF - WAF Release - 2026-02-16
- Published: 2026-02-16T00:00:00+00:00
- Products: Cloudflare
- Tags: security, feature, platform
- URL: https://standarddb.com/cloudflare/cloudflare-waf-adds-detections-for-zimbra-lfi-and-vite-path-traversal--vswjzdjf
- Original: https://developers.cloudflare.com/changelog/2026-02-16-waf-release

## New CVE Detections

Cloudflare has rolled out proactive WAF rules to defend against two newly disclosed vulnerabilities affecting widely-used software:

**CVE-2025-68645 (Zimbra LFI)**: A local file inclusion vulnerability in Zimbra Collaboration Suite versions 10.0 and 10.1 allows unauthenticated attackers to craft malicious requests to the `/h/rest` endpoint. By exploiting improper internal dispatching logic, attackers can include arbitrary files from the WebRoot directory, potentially exposing sensitive configuration, credentials, and user data.

**CVE-2025-31125 (Vite Path Traversal)**: Vite's JavaScript development server exposes non-allowed files when accessed with specific `?inline&import` query parameters on network-exposed instances. This enables unauthorized file reads and can leak sensitive information including source code and environment variables.

## Rule Changes

Two new rules have been added to the **Cloudflare Managed Ruleset** with immediate blocking enabled:

- **Zimbra Rule** (ID: `...833761f7`): Detects Local File Inclusion attempts against the Zimbra `/h/rest` endpoint. Previous action was Log; now set to Block.
- **Vite Rule** (ID: `...950ed8c8`): Detects WASM import path traversal attempts against Vite servers. Previous action was Log; now set to Block.

## Action Items

- Existing WAF-enabled domains will automatically receive these protections
- Review your WAF logs to identify any legitimate traffic affected by the new blocking rules
- Consider upgrading vulnerable Zimbra instances (ZCS 10.0/10.1) and ensure Vite development servers are not exposed to untrusted networks

---

### v0.16.2-rc0: mlxrunner fixes (#14247)
- Published: 2026-02-14T06:30:42+00:00
- Products: 
- Tags: release, bugfix
- URL: https://standarddb.com/ollama/ollama-v0-16-2-rc0-fixes-mlxrunner-model-loading-and-image-generation--ysztnrxp
- Original: https://github.com/ollama/ollama/releases/tag/v0.16.2-rc0

## Overview

Ollama v0.16.2-rc0 is a release candidate focusing on bug fixes for the mlxrunner component, which handles model execution on Apple Silicon and other platforms.

## Changes

- **Model Loading**: Fixed loading of GLM4 MOE Lite models from the mlxrunner
- **Diffusion Models**: Resolved issues preventing diffusion models from loading correctly
- **Image Generation**: Fixed the `--imagegen` flag to properly enable image generation features
- **Code Cleanup**: Removed unnecessary log lines to reduce noise in output

## What Developers Need to Know

If you've experienced issues with:
- Loading specific models like GLM4 MOE Lite
- Running diffusion models on Apple Silicon
- Using the `--imagegen` flag for image generation tasks

This release candidate addresses these problems. Test this RC version and provide feedback before the stable v0.16.2 release.

---

### Shedding old code with ecdysis: graceful restarts for Rust services at Cloudflare
- Published: 2026-02-13T14:00:00+00:00
- Products: Cloudflare
- Tags: open-source, feature, sdk, release
- URL: https://standarddb.com/cloudflare/cloudflare-open-sources-ecdysis-a-rust-library-for-zero-downtime-service--qpfzmhtp
- Original: https://blog.cloudflare.com/ecdysis-rust-graceful-restarts

## Graceful Restarts at Scale

Cloudflare operates network services handling millions of requests per second globally, making downtime—even brief—catastrophic. Critical services performing traffic routing, TLS lifecycle management, and firewall enforcement cannot afford interruptions. When updates and security patches are needed, they must deploy without dropping connections or failing requests.

## The Problem

The naive approach of stopping the old process and starting a new one creates a window where incoming connections are refused with `ECONNREFUSED`. For services handling thousands of requests per second across hundreds of data centers, a 100ms gap translates to hundreds of dropped connections. Additionally, stopping the old process immediately terminates all established connections—abruptly disconnecting clients uploading files, streaming video, or maintaining WebSockets and gRPC streams.

Attempts to use `SO_REUSEPORT` to bind multiple processes to the same port before shutdown fail because the kernel assigns incoming connections to listening sockets but terminates orphaned connections if a process exits before accepting them.

## How ecdysis Works

ecdysis uses an approach pioneered by NGINX:

- **The parent process forks a child process**, which replaces itself with the new code version via `execve()`
- **Socket file descriptors are inherited** via a named pipe shared with the parent
- **Both processes temporarily share the listening socket**, allowing the parent to continue accepting connections during the child's initialization
- **The child signals readiness** to the parent, which then closes its socket copy and drains remaining connections
- **If the child crashes during initialization**, the parent never stopped listening, so no connections are dropped

This architecture eliminates coverage gaps while providing a safe initialization window for new code.

## Features and Integration

ecdysis provides native support for modern Rust development:

- **Tokio integration**: Async stream wrappers allow inherited sockets to become listeners without additional glue code
- **systemd support**: Optional systemd-notify integration for process lifecycle management
- **Crash safety**: Failed initializations don't interrupt the running service

## Availability

The library is now available on GitHub, crates.io, and docs.rs under an open-source license, enabling any organization running critical Rust services to implement zero-downtime upgrades.

---

### Stale-if-error cache-control directive now supported for all responses
- Published: 2026-02-13T13:00:00+00:00
- Products: Vercel
- Tags: feature, api, performance
- URL: https://standarddb.com/vercel/vercel-cdn-adds-stale-if-error-cache-directive-for-resilient-error-handling--pnfptmzg
- Original: https://vercel.com/changelog/stale-if-error-cache-control-header-is-now-supported

## Stale-if-Error Caching Now Available

Vercel has expanded its Cache-Control directive support to include `stale-if-error`, giving developers more control over how their applications handle origin failures. When configured, this directive instructs the CDN to serve a previously cached response instead of returning an error to the client when the origin becomes unavailable.

## How It Works

The `stale-if-error` directive specifies a time window (in seconds) during which stale cached content can be served when origin requests fail. Common failure scenarios include:

- **5xx Server Errors** (e.g., 500 Internal Server Error)
- **Network Failures** and connectivity issues
- **DNS Resolution Errors**

When the origin returns an error response and the directive is present, the CDN may serve the most recent cached version of the resource instead of propagating the error, allowing your application to continue functioning gracefully.

## Getting Started

Developers can configure `stale-if-error` by adding it to their Cache-Control headers. For detailed implementation instructions and best practices, refer to the [official documentation](https://vercel.com/docs/headers/cache-control-headers#stale-if-error).

This feature is particularly valuable for applications that depend on multiple upstream services, providing an additional layer of resilience and improving user experience during transient failures.

---

### Introducing Lockdown Mode and Elevated Risk labels in ChatGPT
- Published: 2026-02-13T10:00:00+00:00
- Products: OpenAI API, ChatGPT
- Tags: feature, security, api
- URL: https://standarddb.com/openai/openai-introduces-lockdown-mode-and-elevated-risk-labels-to-mitigate-prompt--qzqhlwwt
- Original: https://openai.com/index/introducing-lockdown-mode-and-elevated-risk-labels-in-chatgpt

## New Security Protections Against Prompt Injection Attacks

As AI systems take on increasingly complex tasks—particularly those involving web access and connected applications—the security landscape has shifted. One emerging threat has become especially critical: **prompt injection attacks**, where malicious third parties attempt to manipulate AI systems into following unauthorized instructions or exposing sensitive information.

OpenAI is introducing two complementary protections designed to help users and organizations mitigate these risks:

### Lockdown Mode: Advanced Protection for High-Risk Users

**Lockdown Mode** is an optional, advanced security setting designed for high-security users such as executives and security teams at prominent organizations. This mode tightly constrains how ChatGPT can interact with external systems to dramatically reduce prompt injection-based data exfiltration risks.

Key features of Lockdown Mode include:

- **Restricted web browsing**: Limited to cached content only—no live network requests leave OpenAI's controlled infrastructure, preventing sensitive data exfiltration through browsing attacks
- **Deterministic feature disablement**: Tools and capabilities that could be exploited by attackers are disabled entirely when strong data safety guarantees cannot be provided
- **Admin-controlled granularity**: Workspace admins can selectively enable or disable specific apps and actions within those apps for users in Lockdown Mode
- **Availability**: Currently available for ChatGPT Enterprise, ChatGPT Edu, ChatGPT for Healthcare, and ChatGPT for Teachers; consumer availability is planned for the coming months

Admins can enable Lockdown Mode through Workspace Settings by creating a new role with these restrictions layered on top of existing security controls.

### Elevated Risk Labels: Transparent Risk Communication

To help users make informed decisions about feature usage, OpenAI is standardizing how it labels capabilities that introduce additional security risks. These **"Elevated Risk" labels** will appear consistently across ChatGPT, ChatGPT Atlas, and Codex, ensuring users receive the same clear guidance regardless of where they encounter these features.

The labels include explanations of what changes, what risks may be introduced, and when that access is appropriate. For example, in Codex (OpenAI's coding assistant), developers can grant network access for tasks like looking up documentation, with the "Elevated Risk" label clearly displayed alongside configuration options.

### Broader Security Strategy

These protections build on OpenAI's existing multi-layered approach, including sandboxing, protections against URL-based data exfiltration, monitoring and enforcement, and enterprise-grade controls like role-based access and audit logs. The **Compliance API Logs Platform** provides detailed visibility into app usage, shared data, and connected sources to maintain admin oversight.

As OpenAI strengthens safeguards for these features, the "Elevated Risk" label will be removed once security advances sufficiently mitigate risks for general use.

---

### Beyond rate limits: scaling access to Codex and Sora
- Published: 2026-02-13T09:00:00+00:00
- Products: OpenAI API, OpenAI
- Tags: feature, api, pricing
- URL: https://standarddb.com/openai/openai-introduces-credit-fallback-system-for-codex-and-sora-users-can-exceed--nghzjjdx
- Original: https://openai.com/index/beyond-rate-limits

## Rethinking Access Control for High-Demand Products

As Codex and Sora adoption accelerated over the past year, OpenAI encountered a common scaling challenge: traditional access models—strict rate limits or immediate usage-based billing—fell short. Rate limits left users with a frustrating "come back later" experience, while pure usage-based billing introduced lag, reconciliation issues, and required charging from the first token without supporting early exploration.

## The Hybrid Access Model

OpenAI built a unified system that combines both approaches. Users operate within a **rate-limit waterfall** that flows through multiple layers: enforced rate limits, free tiers, promotional access, enterprise entitlements, and finally credits. When a user exceeds their rate limit, the system seamlessly transitions to credit consumption in the same request, making the credit system feel invisible to end users.

The key architectural shift was modeling access as a **decision waterfall rather than a binary gate**. Instead of asking "is this allowed?", the system asks "how much is allowed, and from where?" This reflects actual user experience—there's no context switch between systems; users simply keep using the product.

## Real-Time Correctness and Auditability

Rather than adopting third-party billing platforms, OpenAI built an in-house distributed system to meet two critical requirements:

- **Real-time decisions:** When a user hits a rate limit, the system must immediately know if credits are available, preventing surprise blocks and balance inconsistencies
- **Full transparency:** Every request includes detailed reasoning: why it was allowed/blocked, usage consumed, and which limits or balances applied

The system maintains three separate, independently auditable datasets: product usage events (what users actually did), monetization events (what gets charged), and balance updates (credit adjustments). This separation enables offline reconciliation and prevents double-charging through idempotent transactions with stable keys.

## Design for Trust

Notably, OpenAI prioritized **provable correctness over strict enforcement**. Credit balance updates are asynchronous (though near-real-time) to create an audit trail, allowing the company to demonstrate the system is functioning correctly. When delays cause users to slightly exceed their balance, the system automatically refunds the overage rather than blocking requests. This design choice reflects that user trust matters more than perfectly strict enforcement.

All balance decrements and update records are committed atomically per account with serialized concurrent request handling, preventing race conditions on shared balances. This foundation now enables OpenAI to offer additional access tiers and flexible monetization for both products.

---

### Scaling social science research
- Published: 2026-02-13T09:00:00+00:00
- Products: OpenAI API, OpenAI
- Tags: open-source, feature, release, api
- URL: https://standarddb.com/openai/openai-releases-gabriel-an-open-source-toolkit-for-quantifying-qualitative--mvfglsqx
- Original: https://openai.com/index/scaling-social-science-research

## What is GABRIEL?

GABRIEL is an open-source Python toolkit designed to help economists, social scientists, and data scientists systematically analyze qualitative data. The tool leverages GPT to transform unstructured text and images into quantitative measurements, addressing a major bottleneck in social science research: the time-consuming manual process of converting rich qualitative data into analyzable, quantitative form.

## How it works

Researchers describe what they want to measure using everyday language (e.g., "how family-friendly is this job listing?"), and GABRIEL automatically applies that same question across thousands or millions of documents, returning scores for each one. This shift from manual labeling to automated, consistent measurement lets researchers focus on higher-value work: choosing what to measure, validating results, and drawing conclusions.

## Use cases and capabilities

GABRIEL can analyze scientific papers to track methodological trends over time, measure how course curricula allocate attention to different subjects, extract historical details from town records across Europe, and identify patterns in customer reviews. Beyond measurement, the toolkit provides practical research tools including dataset merging (even with mismatched columns), deduplication, passage coding, theory generation assistance, and text deidentification to preserve privacy.

## Availability and next steps

The toolkit is now available as an open-source Python library on GitHub with a tutorial notebook. It's designed to require minimal technical background. Researchers can get started immediately and provide feedback to drive future improvements. The accompanying research paper benchmarks GPT's accuracy at labeling qualitative data across diverse use cases, demonstrating its effectiveness for research applications.

---

### v0.5.89
- Published: 2026-02-13T08:54:25+00:00
- Products: Sim Studio
- Tags: release, feature, integration, bugfix, platform
- URL: https://standarddb.com/sim/sim-studio-v0-5-89-adds-google-books-integration-and-opus-4-5-support--kgwlrbnm
- Original: https://github.com/simstudioai/sim/releases/tag/v0.5.89

## New Features

**Google Books Integration**: Sim Studio now supports integration with Google Books, enabling workflows to interact with book data and library functionality.

**Creator Tools**: Added comprehensive creator monetization and tracking features including referrer tracking, code redemption, and campaign tracking capabilities for content creators using the platform.

**Internal API Improvements**: Enhanced internal API infrastructure with dedicated base URLs for internal service calls.

## Notable Improvements and Fixes

**Model Updates**: The Copilot default model has been updated to Opus 4.5, providing improved performance for AI-powered workflows.

**Tool Input System**: Refactored the tool-input component with subblock-first rendering and extracted components for better performance and maintainability. Fixed synchronization of cleared subblock values to tool parameters.

**Terminal & Agent Stability**: 
- Fixed terminal reconnection logic to restore sessions after page refresh
- Improved custom tool handling to always fetch the latest version from the database when a custom tool ID is specified

**S3 Storage**: Enhanced S3 integration with region override support for get-object operations and more robust URL parsing.

**Change Detection**: Resolved false positive triggers in the change detection system for workflow blocks.

## Technical Details

The release includes several internal refactoring efforts, including migration of response handling to TypeScript types and improved error handling across multiple systems.

---

### Olmix: A framework for data mixing throughout LM development
- Published: 2026-02-13T08:00:00+00:00
- Products: 
- Tags: open-source, feature, platform
- URL: https://standarddb.com/ai2/olmix-framework-enables-efficient-data-mixing-for-language-model-development--kgkglcqh
- Original: https://allenai.org/blog/olmix

## Olmix: Solving Data Mixing at Scale

Modern language models are trained on heterogeneous data sources, but determining the optimal mix of these sources has traditionally relied on manual experimentation and guesswork. AI2's new **Olmix** framework addresses two persistent challenges in LM development:

1. **Lack of configuration guidance**: Existing mixing approaches make inconsistent choices about model sizes, experiment counts, and regression techniques—with little concrete justification in the literature.
2. **Expensive recomputation**: As training datasets evolve through development iterations, practitioners typically must recompute mixtures from scratch, becoming a computational bottleneck.

## OlmixBase: Empirically Grounded Configuration

The framework's first component, **OlmixBase**, provides research-validated defaults through systematic experimentation:

- **Proxy model size**: 15M parameters strike the optimal balance—large enough to achieve >0.89 rank correlation with target models, small enough to remain computationally efficient. Substantially smaller proxies (1M parameters) produce unreliable results (ρ = 0.73).
- **Mixing cost scaling**: Required proxy runs scale linearly with domain count (O(m) runs for m domains), giving practitioners a concrete compute allocation formula.
- **Regression model**: Log-linear regression provides consistent performance across settings while remaining competitive on downstream validation, replacing the ad-hoc model selection common in prior work.
- **Data feasibility constraints**: The method incorporates explicit repetition limits to prevent over-allocating scarce data types.

## Mixture Reuse for Iterative Development

The second component introduces **mixture reuse**—a technique for efficiently updating mixtures when datasets change. Rather than recomputing from scratch, unchanged domains are bundled into a "virtual domain" while only changed domains are recomputed. This approach reduces compute costs significantly during the iterative dataset refinement typical of real LM development.

## Empirical Results

In experiments on DCLM data across 52 downstream tasks (math, code, commonsense QA), Olmix achieved a **12% improvement** on downstream tasks and demonstrated **3x better data efficiency** compared to no mixing approach.

The framework is available as [open-source code](https://github.com/allenai/olmix) with accompanying [technical report](https://arxiv.org/abs/2602.12237).

---

### v0.16.1
- Published: 2026-02-13T01:18:29+00:00
- Products: 
- Tags: release, bugfix
- URL: https://standarddb.com/ollama/ollama-v0-16-1-improves-installer-ux-and-image-generation-timeouts--bfblvszt
- Original: https://github.com/ollama/ollama/releases/tag/v0.16.1

## Installation Experience Improvements

Ollama v0.16.1 focuses on improving the user experience during installation and configuration:

**macOS Installation**: The `curl` install script will now intelligently prompt for your password only when required, reducing friction during setup.

**Windows Installation**: The Windows installer (`iex` script) now displays progress feedback during installation, providing better visibility into the setup process.

## Image Generation Timeout Support

Image generation models will now properly respect the `OLLAMA_LOAD_TIMEOUT` environment variable. This allows users to customize how long the system waits for image generation models to load before timing out, improving flexibility in resource-constrained or slower environments.

## Getting Started

Download the latest release from the [Ollama GitHub releases page](https://github.com/ollama/ollama/releases/tag/v0.16.1) for your platform (macOS, Windows, or Linux).

---

### Show profile names (first and last name) alongside user handles
- Published: 2026-02-13T00:00:00+00:00
- Products: GitHub
- Tags: feature, platform
- URL: https://standarddb.com/github/github-adds-configurable-profile-name-visibility-for-enterprise-and--jpjylphs
- Original: https://github.blog/changelog/2026-02-13-you-can-now-show-profile-names-alongside-user-handles

## Profile Name Visibility Control Now Available

GitHub has released a new configuration option that allows enterprise and organization administrators to control the visibility of profile names (first and last names) alongside user handles across the platform.

### Where This Applies

The setting applies across all GitHub spaces, including:
- Repositories
- Issues
- Pull requests
- Discussions
- Public, private, and internal repositories within your enterprise organizations

### Configuration Options

**Enterprise-level settings** provide three choices:
- **Enable everywhere**: Profile names appear across all organizations
- **Disable everywhere**: Profile names are hidden across all organizations
- **Let organizations decide**: Individual organizations can configure their own preference

**Organization-level settings** can only be configured if the enterprise policy is set to "Let organizations decide". Organizations can then toggle the feature on or off independently.

### How to Configure

Admins can access these settings by navigating to their enterprise or organization Settings, then finding the "Member appearance" section under "Profile name visibility." GitHub has also published detailed documentation in their enterprise account guide for additional guidance.

This feature gives enterprises more granular control over how user identities are displayed, balancing transparency with privacy and accessibility preferences across different organizational contexts.

---

### Workers VPC - Origin CA certificate support for Workers VPC
- Published: 2026-02-13T00:00:00+00:00
- Products: Cloudflare Workers
- Tags: feature, api, security
- URL: https://standarddb.com/cloudflare/workers-vpc-now-supports-cloudflare-origin-ca-certificates-for-https-connections--fmykrbtb
- Original: https://developers.cloudflare.com/changelog/2026-02-13-origin-ca-certificate-support

## New Certificate Support

Workers VPC now supports Cloudflare Origin CA certificates when connecting to your private services over HTTPS. This expands the types of TLS certificates that Workers VPC trusts beyond publicly issued certificates.

## Previous Limitations

Prior to this update, Workers VPC only trusted certificates issued by publicly trusted certificate authorities such as Let's Encrypt and DigiCert. This required developers to provision certificates from external public CAs even for internal, private network connections.

## Key Benefits

- **Cost savings**: Use free Cloudflare Origin CA certificates instead of provisioning from public CAs
- **Simplified setup**: Encrypt traffic between the tunnel and your service without external CA dependencies
- **Enhanced security**: Maintain HTTPS encryption for internal connections within private networks
- **Native integration**: Leverage Cloudflare's certificate infrastructure already available to your account

## How to Use

Deploy Cloudflare Origin CA certificates on your origin servers within private networks and connect to them from Workers VPC using the standard `https` scheme. For detailed configuration instructions, refer to the [Supported TLS certificates documentation](https://developers.cloudflare.com/workers-vpc/configuration/vpc-services/#supported-tls-certificates).

---

### Workers, Agents, Workers AI - Introducing GLM-4.7-Flash on Workers AI, @cloudflare/tanstack-ai, and workers-ai-provider v3.1.1
- Published: 2026-02-13T00:00:00+00:00
- Products: Cloudflare Workers, Cloudflare
- Tags: release, feature, model, api, sdk, integration
- URL: https://standarddb.com/cloudflare/cloudflare-adds-glm-4-7-flash-model-and-tanstack-ai-support-to-workers-ai--jsctlvhc
- Original: https://developers.cloudflare.com/changelog/2026-02-13-glm-47-flash-workers-ai

## GLM-4.7-Flash: New Multilingual Model for Workers AI

Cloudflare has introduced **GLM-4.7-Flash**, a fast and efficient multilingual text generation model optimized for dialogue, instruction-following, and agent applications. The model features a 131,072 token context window, making it suitable for long-form content generation, complex reasoning, and processing lengthy documents.

**Key capabilities include:**
- Multi-turn tool calling for building AI agents that invoke functions across conversation turns
- Native multilingual support for content generation across multiple languages
- Fast inference optimized for low-latency chatbots and virtual assistants
- Excellent instruction-following for code generation and structured tasks

Access the model through Workers AI bindings (`env.AI.run()`), REST endpoints (`/run` or `/v1/chat/completions`), AI Gateway, or the Vercel AI SDK via the workers-ai-provider.

## TanStack AI Integration

The new **@cloudflare/tanstack-ai** package brings Workers AI and AI Gateway support to TanStack AI, offering a framework-agnostic alternative for developers preferring TanStack's approach. The adapters support four configuration modes: plain binding, plain REST, AI Gateway binding, and AI Gateway REST.

**Available adapters:**
- **Chat** (`createWorkersAiChat`) — Streaming completions with tool calling and structured output
- **Image generation** (`createWorkersAiImage`) — Text-to-image models
- **Transcription** (`createWorkersAiTranscription`) — Speech-to-text processing
- **Text-to-speech** (`createWorkersAiTts`) — Audio generation
- **Summarization** (`createWorkersAiSummarize`) — Text summarization

AI Gateway adapters also route requests from OpenAI, Anthropic, Gemini, Grok, and OpenRouter through Cloudflare's infrastructure for caching, rate limiting, and unified billing.

## Enhanced Vercel AI SDK Support

The **workers-ai-provider v3.1.1** release adds three new capabilities to the Vercel AI SDK integration: **transcription** (speech-to-text), **text-to-speech** (audio generation), and **reranking** (document reordering for RAG pipelines and search results).

The update also includes a comprehensive reliability overhaul addressing streaming, tool calling, and premature stream termination. Streams now process token-by-token with proper backpressure handling, tool call ID sanitization has been fixed, and unexpected stream terminations now correctly report error status instead of silently succeeding.

**Getting started:**
```bash
npm install @cloudflare/tanstack-ai @tanstack/ai
npm install workers-ai-provider@latest ai
```

All three updates enable developers to build production-grade AI applications that run entirely on Cloudflare's edge network with framework flexibility and expanded multimodal capabilities.

---

### New features and improvements in GitHub Copilot in JetBrains IDEs
- Published: 2026-02-13T00:00:00+00:00
- Products: GitHub Copilot
- Tags: feature, release, platform
- URL: https://standarddb.com/github/github-copilot-in-jetbrains-ides-gains-agent-skills-in-public-preview--krwwtgnh
- Original: https://github.blog/changelog/2026-02-13-new-features-and-improvements-in-github-copilot-in-jetbrains-ides-2

## Agent Skills Support in Public Preview

GitHub Copilot for JetBrains IDEs now supports **Agent Skills**, a new feature that enables developers to tailor Copilot for their specific workflows. Skills help reduce repeated setup and automatically load skill-specific content into context when needed. Developers can create custom skills for their projects or leverage community-created skills from repositories like [github/awesome-copilot](https://github.com/github/awesome-copilot) and [anthropics/skills](https://github.com/anthropics/skills).

To enable Agent Skills, navigate to **Settings > GitHub Copilot > Chat > Agent**. For Copilot Business or Enterprise subscribers, administrators must first enable the **Editor preview features** policy.

## Enhanced User Experience

The update introduces several quality-of-life improvements across the IDE:

- **Inline Chat Enhancements**: Quickly add selected code as contextual input to Copilot Chat, and open inline chat directly from the floating code toolbar for faster iteration.
- **Refined Settings Management**: Individual toggles now allow you to independently control Agent mode, Coding Agent, and Custom Agent. A sign-in reminder appears if you're not authenticated.
- **Improved Navigation**: Smoother file navigation during collapse/expand operations, more intuitive prompt navigation with Home and End keys, and redesigned chat and diff views for better readability.

## Quality and Stability Improvements

The release prioritizes reliability with enhanced safety checks in file-handling tools, increased stability for next edit suggestions, and persistent mode settings that survive IDE restarts.

## Getting Started

Update to the [latest version of the GitHub Copilot plugin](https://plugins.jetbrains.com/plugin/17718-github-copilot--your-ai-pair-programmer/versions) to access these features. Share feedback via in-product feedback options or the [GitHub Copilot for JetBrains IDEs feedback repository](https://github.com/microsoft/copilot-intellij-feedback/issues).

---

### New repository settings for configuring pull request access
- Published: 2026-02-13T00:00:00+00:00
- Products: GitHub
- Tags: feature, platform
- URL: https://standarddb.com/github/github-adds-pull-request-access-controls-allowing-maintainers-to-disable-or--nftsctzs
- Original: https://github.blog/changelog/2026-02-13-new-repository-settings-for-configuring-pull-request-access

## Two New Pull Request Controls

GitHub maintainers now have granular control over how repositories accept contributions through two new repository settings available in **Settings > General > Features**:

### Disable Pull Requests Entirely

Repositories can now turn off pull requests completely, similar to existing controls for wikis, issues, discussions, and projects. When disabled:
- The pull requests tab becomes invisible
- No one can view existing pull requests or open new ones
- Useful for mirror repositories, read-only codebases, and projects shared publicly without accepting contributions

### Restrict to Collaborators Only

A new setting allows pull request creation exclusively from collaborators (users with write access) while keeping the pull requests tab visible. This enables:
- All pull requests remain viewable and commentable
- Only collaborators can open new pull requests
- Fine-grained control during critical development phases

Collaborators can be managed through the repository's **Collaborators** tab in settings.

## Availability and Implementation

These settings are immediately available for all public and private repositories. Full mobile app support is coming soon—currently, disabled pull requests still show the tab but prevent new submissions.

Maintainers can also continue using [temporary interaction limits](https://docs.github.com/en/communities/moderating-comments-and-conversations/limiting-interactions-in-your-repository) to restrict activity from specific users on public repositories.

For additional context, see GitHub's broader maintainer support [blog post](https://github.blog/open-source/maintainers/welcome-to-the-eternal-september-of-open-source-heres-what-we-plan-to-do-for-maintainers/).

---

### GitHub Agentic Workflows are now in technical preview
- Published: 2026-02-13T00:00:00+00:00
- Products: GitHub Copilot, GitHub Actions, GitHub
- Tags: release, feature, api, platform, integration
- URL: https://standarddb.com/github/github-launches-agentic-workflows-in-technical-preview-enabling-ai-driven--qrmkqytl
- Original: https://github.blog/changelog/2026-02-13-github-agentic-workflows-are-now-in-technical-preview

## Intent-Driven Automation in Markdown

GitHub Agentic Workflows simplify repository automation by replacing complex YAML configuration with natural language Markdown descriptions. Instead of hand-crafting GitHub Actions workflows, developers can now describe their automation goals in plain English, and AI agents intelligently convert these descriptions into executable workflows.

## How It Works

Developers add Markdown files to `.github/workflows/` that describe what they want to automate. The `gh aw` CLI extension compiles these natural language descriptions into standard GitHub Actions workflows. These workflows execute using GitHub Copilot CLI or other compatible AI coding agents, with support for multiple triggers including issue/PR events, schedules, manual dispatch, and comment-based invocation.

## Security-First Architecture

The platform prioritizes security with several built-in safeguards:
- **Read-only by default**: Workflows have minimal permissions out of the box
- **Sandboxed execution** with network isolation
- **SHA-pinned dependencies** to prevent supply chain attacks
- **Sanitized write operations** through preapproved "safe outputs"

## Capabilities and Integration

Agentic Workflows come with deep GitHub platform integration through the GitHub MCP Server, providing native access to repositories, issues, pull requests, actions, and security features. Additional tools include browser automation, web search, and support for custom MCP integrations.

Developers can create, edit, debug, and optimize workflows using AI agents directly in VS Code, github.com, or their preferred coding environment. The entire implementation is open source under the MIT license at [`github/gh-aw`](https://github.com/github/gh-aw).

## Getting Started

The feature is available immediately through the `gh aw` CLI extension. GitHub provides comprehensive documentation and reference implementations through Peli's Agent Factory, which showcases over 50 specialized workflows for common use cases like issue triage, CI failure investigation, documentation maintenance, and test coverage improvement.

---

### Chat with your Canvas, DB Metrics, Network flows
- Published: 2026-02-13T00:00:00+00:00
- Products: 
- Tags: feature, platform, api
- URL: https://standarddb.com/railway/railway-releases-ai-agent-for-canvas-postgres-metrics-and-network-flow--xhxwzfsr
- Original: https://railway.com/changelog/2026-02-13-chat-with-your-canvas

## Canvas Chat Agent

Railway has introduced a conversational AI agent that lets developers interact with their infrastructure through natural language. Available first in Priority Boarding, the agent can stage changes to environments, report live status, debug failing deploys, diagnose application issues, and provide recommendations for Railway Functions. Users can access it at dev.new or via the chat icon in existing projects after enabling the feature flag.

## Postgres Metrics and Observability

Postgres databases on Railway now offer dedicated metrics views including connection breakdowns (active, idle, idle in transaction), cache hit ratios, and table-level statistics. The standout feature is the Query Statistics panel, which surfaces expensive queries with call counts, rows returned, and timing percentiles—helping developers identify performance bottlenecks without external tools.

## Network Flow Visualization

The canvas now displays live network traffic between services as animated connections, with line thickness indicating throughput. Network flows show layer 4 data aggregated every 5 seconds, including traffic source, destination, speed, and packet loss. The new Network Flow Logs tab provides detailed protocol, address, volume, latency, and status information for each service.

## Additional Updates

- **CLI Telemetry**: Railway shipped opt-in telemetry in its CLI to improve developer experience. No source code or environment variables are collected. Developers can opt out with `DO_NOT_TRACK=1` or `RAILWAY_NO_TELEMETRY=1`.
- **API Token Visibility**: Developers can now view the scope level for API tokens to quickly understand access permissions.

---

### Network configuration changes for Copilot coding agent
- Published: 2026-02-13T00:00:00+00:00
- Products: GitHub Copilot, GitHub Actions, GitHub
- Tags: feature, api, breaking-change
- URL: https://standarddb.com/github/github-copilot-coding-agent-switches-to-plan-based-network-routing-on-feb-27--ctrgkptv
- Original: https://github.blog/changelog/2026-02-13-network-configuration-changes-for-copilot-coding-agent

## What's Changing

Starting February 27, 2026 at 00:00 UTC, GitHub Copilot coding agent will transition from connecting to a single endpoint (`api.githubcopilot.com`) to **subscription-based network routing** that directs traffic to different hosts depending on the initiating user's Copilot plan:

- **Copilot Business**: `api.business.githubcopilot.com`
- **Copilot Enterprise**: `api.enterprise.githubcopilot.com`
- **Copilot Pro and Pro+**: `api.individual.githubcopilot.com`

This change affects teams who have configured the Copilot coding agent to run on **self-hosted runners** or **larger runners with Azure private networking**. If your organization falls into this category and has run at least one agent session in the past 60 days, GitHub has already sent you an email notification with a list of potentially-affected repositories.

## Action Items Required

You must update your network configurations by the February 27 deadline to maintain continuity:

1. **Identify affected repositories** by checking the `runs-on` value in `.github/workflows/copilot-setup-steps.yml` files in your repositories. If this file doesn't exist, that repository is not affected.

2. **Update firewall/network rules** to allow connections to the new subscription-based endpoints listed above. You may need to allow multiple hosts if your organization has users on different Copilot plans.

3. **Test connectivity** before the February 27 deadline to ensure the Copilot agent runtime can communicate with GitHub systems.

## Important Deadline

If network configuration updates are not completed by February 27, 2026, Copilot coding agent tasks will fail because the agent runtime will be unable to reach the new endpoints.

---

### Custom Kernels for All from Codex and Claude
- Published: 2026-02-13T00:00:00+00:00
- Products: 
- Tags: feature, sdk, integration, platform
- URL: https://standarddb.com/huggingface/hugging-face-releases-cuda-kernel-skill-for-claude-and-codex-agents--tcttrtsc
- Original: https://huggingface.co/blog/custom-cuda-kernels-agent-skills

## New Agent Skill for CUDA Kernel Development

Hugging Face has released a new agent skill that enables coding agents to automatically write optimized CUDA kernels for machine learning workloads. The skill teaches agents how to generate production-grade kernels complete with PyTorch bindings, GPU-specific optimizations, and comprehensive benchmarking—addressing a significant gap where domain expertise was previously scattered across documentation and Stack Overflow.

### What's Included

The skill ships with the `kernels` library and can be installed via a single command (`kernels skills add cuda-kernels --claude`). It consists of approximately 550 tokens of structured guidance plus reference scripts, GPU optimization guides, and working examples covering:

- **Hardware optimization** for NVIDIA GPUs (H100, A100, T4) including compute capabilities, memory bandwidth profiles, and block sizing strategies
- **Integration patterns** for both `diffusers` and `transformers` libraries, with library-specific pitfalls and solutions
- **Kernel templates** with vectorized memory access patterns for BF16, FP16, and FP32 data types
- **Benchmarking workflows** for isolated micro-benchmarks and end-to-end pipeline performance validation
- **Kernel Hub integration** for loading pre-compiled community kernels

### How Developers Use It

Once installed, developers can prompt their coding agent with specific requests:

```
Build a vectorized RMSNorm kernel for H100 targeting the Qwen3-8B model in transformers.
```

The agent reads the skill, selects appropriate architecture parameters, generates CUDA source code, writes PyTorch bindings, configures `build.toml`, and creates benchmark scripts—all without manual intervention.

### Supported Agents

The skill supports Claude Code, Cursor, Codex, and OpenCode, with options to install globally or to custom destinations. Hugging Face is also open to contributions to the skill itself via GitHub.

### Context and Impact

This release addresses a critical pain point in CUDA kernel development: the vast surface area of domain knowledge required, from hardware-specific optimization guides to library integration patterns. By packaging this expertise into an agent skill, Hugging Face makes it possible for agents to produce working, optimized kernels end-to-end. The authors demonstrated success with real production targets including both diffusers pipelines and transformers models.

---

### Advanced filters and share issues in private teams
- Published: 2026-02-13T00:00:00+00:00
- Products: Linear
- Tags: release, feature, api, integration, performance, bugfix
- URL: https://standarddb.com/linear/linear-adds-advanced-filters-and-private-team-issue-sharing-in-february-release--hxccgpww
- Original: https://linear.app/changelog/2026-02-13-advanced-filters-and-share-issues-in-private-teams

## Advanced Filters for Precision Views

Linear now supports advanced filters with **multiple AND/OR conditions**, allowing users to combine fields like Priority, Labels, and Customer status in a single view. This enables teams to track specific issue sets more effectively—for example, high-priority bugs for prospective customers can be consolidated into one dashboard view.

Users can access advanced filters through the filter menu or use **AI filter** to describe their needs in natural language. Filters can power targeted dashboards and support view subscriptions for notifications when issues match specified conditions.

## Private Team Issue Sharing

Enterprise plan users can now **share individual issues from private teams with external collaborators** without granting broader team access. This is particularly valuable for sensitive teams like security or HR, where specific problems need outside expertise without exposing the entire team's data. Shared issues display a prominent banner indicating visibility scope.

## Mobile and Integration Improvements

The update brings project and initiative creation to Android and iOS, enabling teams to capture ideas on-the-go. A new **Gemini Enterprise connector** lets users create issues and access existing project data directly from Google's workspace.

## Additional Updates

- **Agent**: Linear Agent for Slack can now attach images from conversations to existing entities
- **Performance**: Improved compatibility with browser security extensions
- **API Changes**: Removed deprecated `CustomerNeed.url` field; added automatic video embeds for YouTube and Loom links in markdown
- **MCP Server**: Added project resources and member listing capabilities
- **Bug Fixes**: Resolved issues with GitHub sync, editor changes, attachment handling, and Slack notification toggling

---

### Cloudflare Fundamentals, SDK - Cloudflare Python SDK v5.0.0-beta.1 now available
- Published: 2026-02-13T00:00:00+00:00
- Products: Cloudflare Workers, Cloudflare
- Tags: sdk, release, breaking-change, api, feature
- URL: https://standarddb.com/cloudflare/cloudflare-python-sdk-v5-0-0-beta-1-introduces-major-breaking-changes-and-40--xrxdmhrm
- Original: https://developers.cloudflare.com/changelog/2026-02-13-cloudflare-python-v500-beta1

## Breaking Changes and Migration Required

Cloudflare Python SDK v5.0.0-beta.1 introduces **extensive breaking changes** affecting 15 resource modules. These changes stem primarily from improvements to the underlying OpenAPI definitions and codegen updates. Affected resources include `abusereports`, `acm.totaltls`, `apigateway.configurations`, `cloudforceone.threatevents`, `d1.database`, `intel.indicatorfeeds`, `logpush.edge`, `origintlsclientauth.hostnames`, `queues.consumers`, `radar.bgp`, `rulesets.rules`, `schemavalidation.schemas`, `snippets`, `zerotrust.dlp`, and `zerotrust.networks`.

**Action required**: Developers using the Python SDK must review the [v5 Migration Guide](https://github.com/cloudflare/cloudflare-python/blob/main/docs/v5-migration-guide.md) before upgrading, as downstream code may require updates.

## 40+ New API Resources

The beta release adds comprehensive coverage for new Cloudflare services:

- **AI Features**: `ai.tomarkdown` (markdown conversion), `aigateway.dynamicrouting`, `aigateway.providerconfigs`, `aisearch`, `aisearch.instances`, `aisearch.tokens`, `radar.ai`
- **Security & Compliance**: `abusereports`, `brandprotection.*` (logos, matches, queries), `fraud`, `tokenvalidation.*`
- **Data & Infrastructure**: `d1.database`, `pipelines.*` (sinks, streams), `r2datacatalog.*`, `connectivity.directory`
- **Real-time Services**: `realtimekit.*` (active sessions, analytics, apps, meetings, recordings, webhooks)
- **Organization & IAM**: `organizations`, `organizations.organizationprofile`, `iam.sso`
- **Networking**: `loadbalancers.monitorgroups`, `origintlsclientauth.*`

## Enhanced Request/Response Handling and Type System

The release includes significant improvements to the SDK's foundation:

- **Type System**: Enhanced type inference for Pyright, improved TypedDict support, added missing method argument and response types, and Pydantic pre-2.8.0 compatibility
- **Request Handling**: Proper multipart/form-data support with JSON data, correct header handling to avoid sending unnecessary defaults, and Content-Type header fixes for GET requests
- **Response Parsing**: Better accuracy for discriminated unions, extra field types, metadata handling, and resource name singularization rules

## Current Status and Next Steps

This is a beta release, so additional breaking changes may occur before the v5.0.0 final release. Developers should test the beta version in non-production environments and report any issues or discrepancies via the Cloudflare SDK repository. The full changelog comparing v4.3.1 to v5.0.0-beta.1 is available on GitHub.

---

### Workers, Agents, Workers AI - Introducing GLM-4.7-Flash on Workers AI, @cloudflare/tanstack-ai, and workers-ai-provider v3.1.1
- Published: 2026-02-13T00:00:00+00:00
- Products: Cloudflare Workers, Cloudflare
- Tags: release, feature, model, api, sdk, integration
- URL: https://standarddb.com/cloudflare/cloudflare-launches-glm-4-7-flash-on-workers-ai-with-multi-turn-tool-calling--jmktdzhx
- Original: https://developers.cloudflare.com/changelog/post/2026-02-13-glm-47-flash-workers-ai

## GLM-4.7-Flash Model Available on Workers AI

Cloudflare has introduced **GLM-4.7-Flash**, a fast multilingual text generation model optimized for dialogue and instruction-following tasks. The model features a 131,072 token context window, making it suitable for long-form content, complex reasoning, and processing extended documents. Key capabilities include:

- **Multi-turn tool calling** for building AI agents that invoke functions across multiple conversation turns
- **Multilingual support** for applications requiring content generation in multiple languages
- **Fast inference** optimized for low-latency responses in chatbots and virtual assistants
- **Instruction following** for code generation and structured task completion

The model is accessible via Workers AI binding (`env.AI.run()`), REST API endpoints (`/run` and `/v1/chat/completions`), AI Gateway, or the Vercel AI SDK through workers-ai-provider.

## New TanStack AI Integration Package

Cloudflare released **@cloudflare/tanstack-ai v0.1.1**, a framework-agnostic package bringing Workers AI and AI Gateway support to TanStack AI. This package provides adapters for four configuration modes and supports:

- **Chat completions** with streaming, tool calling, structured output, and reasoning text
- **Image generation** using available text-to-image models
- **Transcription** for speech-to-text conversion
- **Text-to-speech** for audio generation
- **Summarization** for text processing

AI Gateway adapters also route requests from third-party providers (OpenAI, Anthropic, Gemini, Grok, OpenRouter) through Cloudflare for caching, rate limiting, and unified billing.

## Enhanced workers-ai-provider with New Capabilities

The **workers-ai-provider v3.1.1** for the Vercel AI SDK now supports three additional capabilities beyond chat and image generation:

- **Transcription** (`provider.transcription(model)`) — Automatic handling of model-specific speech-to-text inputs
- **Text-to-speech** (`provider.speech(model)`) — Audio generation with voice and speed customization
- **Reranking** (`provider.reranking(model)`) — Document reranking for RAG pipelines and search optimization

A major reliability overhaul (v3.0.5) fixes streaming token-by-token delivery, tool call ID handling, conversation history preservation, and introduces error detection for premature stream termination. The `createAutoRAG` export has been renamed to `createAISearch` with backward compatibility.

## Getting Started

Install the packages via npm:
```
npm install @cloudflare/tanstack-ai @tanstack/ai
npm install workers-ai-provider@latest ai
```

Developers can immediately begin building agentic applications that run entirely at the edge with GLM-4.7-Flash's multi-turn tool calling combined with TanStack AI or Vercel AI SDK integrations.

---

### Workers VPC - Origin CA certificate support for Workers VPC
- Published: 2026-02-13T00:00:00+00:00
- Products: Cloudflare Workers
- Tags: feature, platform, api
- URL: https://standarddb.com/cloudflare/cloudflare-workers-vpc-now-supports-origin-ca-certificates-for-private-https--xjqcdsbn
- Original: https://developers.cloudflare.com/changelog/post/2026-02-13-origin-ca-certificate-support

## Overview

Workers VPC now accepts Cloudflare Origin CA certificates when connecting to private services over HTTPS. This expands the certificate options previously limited to publicly trusted certificate authorities (CAs) such as Let's Encrypt and DigiCert.

## What Changed

Previously, Workers VPC only validated certificates issued by public CAs. You can now use **free Cloudflare Origin CA certificates** on origin servers within private networks and connect to them securely from Workers VPC using the `https` scheme.

## Use Cases & Benefits

- **Cost-effective encryption**: Leverage free Origin CA certificates instead of provisioning certificates from external public CAs
- **Private network security**: Encrypt tunnel-to-service traffic within your private network without public CA requirements
- **Seamless integration**: Use the standard `https` scheme with Origin CA certificates just as you would with other supported CA types

## Action Items

To use this feature, configure your origin servers with Cloudflare Origin CA certificates and reference them when connecting from Workers VPC. Refer to the [Supported TLS certificates documentation](https://developers.cloudflare.com/workers-vpc/configuration/vpc-services/#supported-tls-certificates) for configuration details.

---

### Cloudflare Fundamentals, SDK - Cloudflare Python SDK v5.0.0-beta.1 now available
- Published: 2026-02-13T00:00:00+00:00
- Products: Cloudflare Workers, Cloudflare
- Tags: release, sdk, breaking-change, api
- URL: https://standarddb.com/cloudflare/cloudflare-python-sdk-v5-0-0-beta-1-adds-40-new-api-resources-with-breaking--nczsyrvm
- Original: https://developers.cloudflare.com/changelog/post/2026-02-13-cloudflare-python-v500-beta1

## Major Expansion of API Resources

Cloudflare's Python SDK v5.0.0-beta.1 introduces a substantial number of new API resources and endpoints. Key additions include:

- **AI & Machine Learning**: `ai.tomarkdown`, `aigateway.dynamicrouting`, `aigateway.providerconfigs`, `aisearch` with instances and tokens management
- **Security & Compliance**: New `fraud` detection, `brandprotection` suite, and enhanced `zerotrust.dlp` capabilities
- **Data & Storage**: R2 Data Catalog integration with credentials and namespace management, D1 database resources
- **Real-time Communications**: Complete Real-time Kit API covering active sessions, live streams, meetings, recordings, and webhooks
- **Infrastructure**: New organization management, pipeline/sink/stream configurations, and load balancer monitor groups

## Breaking Changes Require Migration Planning

This is a major version bump with **15+ resources experiencing breaking changes**, including:
- `abusereports`, `acm.totaltls`, `apigateway.configurations`
- `cloudforceone.threatevents`, `d1.database`, `intel.indicatorfeeds`
- `logpush.edge`, `origintlsclientauth` resources, `queues.consumers`
- `radar.bgp`, `rulesets.rules`, `zerotrust.dlp`, and more

These changes stem from OpenAPI schema accuracy improvements and codegen updates. Developers must **consult the [v5 Migration Guide](https://github.com/cloudflare/cloudflare-python/blob/main/docs/v5-migration-guide.md)** before upgrading.

## Enhanced Type System & Request Handling

The release includes significant improvements to type inference, Pydantic compatibility, and request/response handling:
- Better Pyright type inference for TypedDict compatibility
- Corrected multipart/form-data request handling
- Fixed GET request headers (no Content-Type)
- Improved discriminated union parsing in responses
- Enhanced response body model accuracy across the board

## Beta Testing Period

This is a **beta release** currently undergoing stability testing. Users are encouraged to test in non-production environments and report issues on GitHub. A stable v5.0.0 release is expected to follow after feedback incorporation.

---

### v0.16.0
- Published: 2026-02-12T23:38:44+00:00
- Products: 
- Tags: release, feature, model
- URL: https://standarddb.com/ollama/ollama-v0-16-0-adds-glm-5-and-minimax-m2-5-models-introduces-app-launcher--kzkxlkvs
- Original: https://github.com/ollama/ollama/releases/tag/v0.16.0

## New Models

Ollama v0.16.0 brings two significant additions to its model library:

- **GLM-5**: A strong reasoning and agentic model from Z.ai with 744B total parameters (40B active), specifically engineered for complex systems engineering and long-horizon tasks.
- **MiniMax-M2.5**: A new state-of-the-art large language model optimized for real-world productivity and coding applications.

## New `ollama launch` Command

The release introduces a new application launcher that makes it simple to use Ollama models with external applications. Users can now execute commands like `ollama launch pi` to automatically configure and run their preferred applications with Ollama's model infrastructure.

## Additional Improvements

- **MLX Runner Enhancements**: Improved support for GLM-4.7-Flash through optimizations to Ollama's MLX runner
- **Text Editor Integration**: Ctrl+G now enables users to edit text prompts in their preferred text editor while running models

## Getting Started

Download v0.16.0 from the [Ollama releases page](https://github.com/ollama/ollama/releases/tag/v0.16.0). Existing users can update to access the new models and launcher functionality immediately.

---

### Gemini 3 Deep Think: Advancing science, research and engineering
- Published: 2026-02-12T16:15:09+00:00
- Products: Gemini
- Tags: release, feature, model, api
- URL: https://standarddb.com/google/google-upgrades-gemini-3-deep-think-achieves-84-6-on-arc-agi-2-and-gold-medal--plykgxhk
- Original: https://deepmind.google/blog/gemini-3-deep-think-advancing-science-research-and-engineering

## Gemini 3 Deep Think: Major Upgrade for Scientific Reasoning

Google has released a significant upgrade to **Gemini 3 Deep Think**, its specialized reasoning mode built to tackle complex challenges in science, research, and engineering. The new version was developed in close partnership with scientists and researchers to handle problems that lack clear guardrails, single correct solutions, or complete data—common characteristics of real-world research work.

### Availability and Access

- **Gemini app**: Now available to Google AI Ultra subscribers
- **Gemini API**: Available for the first time through early access. Researchers, engineers, and enterprises can express interest via the official sign-up form.

### Benchmark Performance

The upgraded Deep Think achieves impressive results across rigorous academic benchmarks:

- **48.4%** on Humanity's Last Exam (without tools)—a new standard for frontier models
- **84.6%** on ARC-AGI-2, verified by the ARC Prize Foundation
- **Elo rating of 3455** on Codeforces competitive programming challenges
- **Gold-medal level** on International Math Olympiad 2025

### Real-World Applications

Early testers have already demonstrated practical impact:

- **Mathematics research**: A Rutgers mathematician used Deep Think to review a technical mathematics paper and identified a subtle logical flaw previously missed by human peer review.
- **Materials science**: Duke University's Wang Lab optimized crystal growth fabrication methods, successfully designing recipes for thin films exceeding 100 μm—a target previous methods struggled to achieve.
- **Hardware design**: Google engineers accelerated the design process for physical components using Deep Think's reasoning capabilities.

The model combines deep scientific knowledge with engineering utility, moving beyond abstract theory to enable practical applications across diverse domains.

---

### Introducing Markdown for Agents
- Published: 2026-02-12T14:03:00+00:00
- Products: Cloudflare Workers, Cloudflare
- Tags: feature, api, platform, integration
- URL: https://standarddb.com/cloudflare/cloudflare-launches-markdown-for-agents-cutting-token-usage-80-with-automatic--mlnrnslt
- Original: https://blog.cloudflare.com/markdown-for-agents

## Markdown for Agents Now Available

Cloudflare has launched **Markdown for Agents**, a new feature that automatically converts HTML pages to markdown format when AI systems and agents request them. The conversion happens in real-time at Cloudflare's edge network for enabled zones, eliminating the need for agents to parse unstructured HTML or perform client-side conversion.

## How It Works

AI agents and crawlers can request markdown versions of pages by including the `Accept: text/markdown` header in their HTTP requests. Cloudflare detects this preference, fetches the original HTML from the origin, converts it to markdown, and serves it back to the client. The service also returns an `x-markdown-tokens` header indicating the token count of the converted markdown, helping agents estimate context window usage and chunking strategy.

### Key Benefits

- **Massive token savings**: The example blog post tested reduces from 16,180 tokens in HTML to just 3,150 in markdown—an 80% reduction
- **Simplified parsing**: Markdown's explicit structure is naturally suited for AI processing, eliminating intent analysis and conversion overhead
- **Content signal support**: Responses include `Content-Signal` headers indicating content can be used for AI training, search results, and agent input
- **Zero developer friction**: Works out-of-the-box for any site on Cloudflare; no configuration required to support markdown requests

## Getting Started

Developers building AI agents can request markdown versions using standard HTTP content negotiation:

```javascript
const r = await fetch('https://example.com/page', {
  headers: { Accept: 'text/markdown, text/html' }
});
const tokenCount = r.headers.get('x-markdown-tokens');
const markdown = await r.text();
```

Cloudflare has already enabled the feature on its developer documentation and blog. Popular coding agents like Claude Code and OpenCode already send these Accept headers—they'll now receive optimized markdown responses automatically.

## Future Roadmap

Cloudflare plans to offer custom Content Signal policies in future updates, allowing site owners to define granular permissions for how their content can be used by different types of AI systems.

---

### Browserbase joins the Vercel Agent Marketplace
- Published: 2026-02-12T13:00:00+00:00
- Products: Vercel
- Tags: integration, feature, platform
- URL: https://standarddb.com/vercel/browserbase-joins-vercel-marketplace-with-one-click-browser-automation-for-ai--wcqmqfzs
- Original: https://vercel.com/changelog/browserbase-joins-the-vercel-agent-marketplace

## Browserbase Now Available on Vercel Marketplace

Browserbase has launched as a one-click integration on the Vercel Marketplace, enabling teams to run browser automation for AI agents without the operational overhead of managing infrastructure.

### Key Capabilities

The integration provides:

- **Seamless connection** to remote browsers over Chrome DevTools Protocol (CDP)
- **Unified billing** with Vercel infrastructure
- **Support for long-lived, stateful sessions** designed for agent workflows
- **Simple setup** with a single API key for installation and connection
- **Web Bot Auth** for Browserbase, allowing agents to reliably browse Vercel-hosted deployments without security interruption

### Use Cases

This integration enables workflows requiring real browser interaction, including:

- Signing in to dashboards
- Filling out forms
- Navigating dynamic pages
- Interacting with Vercel-hosted applications

### Getting Started

Developers can install Browserbase directly from the Vercel Marketplace and immediately begin building browser-based agent workflows. The integration works seamlessly with Vercel Sandbox and AI Gateway. Example templates are available to help teams see the integration in action.

---

### New deployments with vulnerable versions of the third-party package next-mdx-remote are now blocked by default
- Published: 2026-02-12T13:00:00+00:00
- Products: Next.js, Vercel
- Tags: security, platform, breaking-change
- URL: https://standarddb.com/vercel/vercel-blocks-deployments-with-vulnerable-next-mdx-remote-versions-by-default--jxcfxrgr
- Original: https://vercel.com/changelog/new-deployments-with-vulnerable-versions-of-next-mdx-remote-are-now-blocked-by-default

## Automatic Security Blocking

Vercel has implemented automatic protection against deployments containing vulnerable versions of the third-party package `next-mdx-remote` that are susceptible to [CVE-2026-0969](https://www.cve.org/CVERecord?id=CVE-2026-0969). Any new deployment attempt containing an affected version will now automatically fail.

## What Developers Should Know

- **Immediate impact**: New deployments with vulnerable `next-mdx-remote` versions are blocked by default
- **Recommended action**: Upgrade to a patched version of `next-mdx-remote` regardless of your hosting provider
- **Override option**: The blocking can be disabled by setting the environment variable `DANGEROUSLY_DEPLOY_VULNERABLE_CVE_2026_0969=1` on your Vercel project (not recommended for production)

## Next Steps

Review your project's dependencies to identify if you're using vulnerable versions of `next-mdx-remote`. If your deployments are being blocked, update to the latest patched version and redeploy. For configuration details on disabling this protection (if necessary), see Vercel's [environment variables documentation](https://vercel.com/docs/environment-variables#creating-environment-variables).

---

### Use MiniMax M2.5 on AI Gateway
- Published: 2026-02-12T13:00:00+00:00
- Products: Vercel AI SDK, Vercel
- Tags: feature, api, model, integration
- URL: https://standarddb.com/vercel/vercel-ai-gateway-adds-minimax-m2-5-model-support-for-full-stack-development--kzchclqr
- Original: https://vercel.com/changelog/use-minimax-m2-5-on-ai-gateway

## MiniMax M2.5 Now Available on AI Gateway

Vercel's AI Gateway now supports MiniMax M2.5, a model designed for comprehensive full-stack development workflows. The model excels at planning before implementation, breaking down functions, structure, and UI design before writing code.

### Key Capabilities

M2.5 handles complete development projects across multiple platforms:
- **Web, Android, iOS, Windows, and Mac** development
- Full development lifecycle from system design through code review
- Improved adaptation to unfamiliar codebases compared to M2.1
- More efficient problem-solving with fewer search iterations

### Integration with AI Gateway

Access the model using the Vercel AI SDK by setting the model parameter to `minimax/minimax-m2.5`:

```tsx
import { streamText } from 'ai';

const result = streamText({
  model: 'minimax/minimax-m2.5',
  prompt: 'Your development task...'
});
```

### Benefits

Using M2.5 through AI Gateway provides:
- **Unified API** for calling models with consistent interfaces
- **Cost tracking and usage monitoring** across deployments
- **Built-in observability** for debugging and performance insights
- **Intelligent provider routing** with automatic retries and failover
- **Bring Your Own Key** support for flexibility in provider selection

Developers can explore the model's performance on the [AI Gateway leaderboard](https://vercel.com/ai-gateway/leaderboards) or test it directly in the model playground before deploying.

---

### GPT-5.3-Codex-Spark is now available in Windsurf
- Published: 2026-02-12T12:00:00+00:00
- Products: 
- Tags: feature, model, integration
- URL: https://standarddb.com/windsurf/openai-s-gpt-5-3-codex-spark-now-available-in-windsurf-via-arena-mode--zpzmgdqz
- Original: https://windsurf.com/blog/gpt-5.3-codex-spark

## New Model Available

OpenAI's **GPT-5.3-Codex-Spark** is now available in Windsurf, expanding the model options available to developers and users of the platform.

## Access and Availability

The model is accessible exclusively through Windsurf's **Arena Mode**, specifically in:
- **Fast Arena** battle group
- **Hybrid Arena** battle group

Note that GPT-5.3-Codex-Spark is not yet available in the regular model selector and remains limited to these arena battle groups.

## Getting Started

To access the new model, users need to download the latest version of Windsurf. Once installed, they can experiment with GPT-5.3-Codex-Spark through the designated arena modes.

---

### Introducing GPT-5.3-Codex-Spark
- Published: 2026-02-12T10:00:00+00:00
- Products: OpenAI API, ChatGPT, OpenAI
- Tags: release, feature, model, api, performance
- URL: https://standarddb.com/openai/openai-releases-gpt-5-3-codex-spark-optimized-for-real-time-coding-at-1000--dbryxkxt
- Original: https://openai.com/index/introducing-gpt-5-3-codex-spark

## GPT-5.3-Codex-Spark Launch

OpenAI has released **GPT-5.3-Codex-Spark**, a new ultra-fast coding model optimized for real-time collaboration and interactive development. This research preview marks the first milestone in OpenAI's partnership with Cerebras, announced in January. The model is specifically designed to feel "near-instant" when making targeted code edits, refactoring logic, and iterating on interfaces—enabling developers to see results immediately.

## Performance and Capabilities

Codex-Spark delivers exceptional speed while maintaining strong coding capability:

- **Throughput**: Over 1000 tokens per second on Cerebras' Wafer Scale Engine 3 hardware
- **Context window**: 128k tokens (text-only for this preview)
- **SWE-Bench Pro**: Achieves 50% accuracy on complex software engineering tasks in just 8 minutes (vs. GPT-5.3-Codex at ~18 minutes)
- **Terminal-Bench 2.0**: 58.4% accuracy for agentic command-line tasks

The model complements OpenAI's frontier models by supporting both long-running autonomous tasks and real-time interactive work. Its lightweight default behavior makes minimal, targeted edits without automatically running tests unless requested.

## System-Wide Latency Improvements

Beyond the model itself, OpenAI has implemented infrastructure optimizations that benefit all models:

- **Per-request overhead**: Reduced by 80% through persistent WebSocket connections
- **Per-token overhead**: Reduced by 30%
- **Time-to-first-token**: Reduced by 50% via streamlined response streaming and inference stack rewrites

These improvements come from the new WebSocket default path in Responses API, which will roll out to all models in the coming weeks.

## Availability and Access

**Current availability:**
- Research preview for ChatGPT Pro users in Codex app, CLI, and VS Code extension
- Limited API access for design partners
- Separate rate limits during preview; usage doesn't count toward standard API limits

**Future roadmap:** OpenAI plans to expand access as they optimize performance under real workloads and eventually introduce larger models, longer context windows, and multimodal input capabilities.

## Safety and Infrastructure Notes

Codex-Spark includes the same safety training as mainline models and passed cybersecurity evaluations under OpenAI's Preparedness Framework. The model runs on Cerebras' specialized hardware, which complements—not replaces—GPUs for cost-effective broad inference. The architecture allows combining GPUs and Cerebras for optimal performance on specific workloads.

---

### How researchers are using AutoDiscovery
- Published: 2026-02-12T08:00:00+00:00
- Products: 
- Tags: feature, platform, release
- URL: https://standarddb.com/ai2/ai2-launches-autodiscovery-an-ai-tool-that-autonomously-explores-datasets-and--xpmgnlgy
- Original: https://allenai.org/blog/autodiscovery-impact

## AutoDiscovery: Autonomous Dataset Exploration

AI2 has released **AutoDiscovery**, an AI-powered research tool that fundamentally shifts how scientists approach data exploration. Unlike traditional AI research tools that require researchers to formulate hypotheses first, AutoDiscovery works in reverse: provide it a dataset and it autonomously generates hypotheses, designs statistical experiments, and surfaces novel findings researchers might never have considered.

The system operates overnight to comprehensively explore structured datasets, delivering a curated list of research directions with **reproducible code, statistical results, and clear paths for follow-up investigation**. AutoDiscovery evolved from a published research project with open-source code, refined through collaboration with domain experts across multiple disciplines.

## Real-World Applications

**Oncology:** Dr. Kelly Paulson's team at Providence Swedish Cancer Institute used AutoDiscovery to analyze large clinical and genomic datasets from breast cancer and melanoma research. The tool confirmed expected findings (immune activity importance in melanoma, PI3K pathway relevance in breast cancer) while surfacing unexpected associations—such as strong immune response patterns in melanoma and lymph node spread risk factors in breast cancer—that weren't part of the team's initial hypotheses and are now undergoing validation.

**Marine Ecology:** Researchers at Scripps Institution of Oceanography applied AutoDiscovery to 20+ years of rocky reef monitoring data from the Gulf of California. Beyond documenting known heatwave impacts, the system discovered mechanistic relationships between productivity across trophic levels that would have required extensive manual iteration to uncover. Researchers emphasized AutoDiscovery's transparency and interpretability as critical for building trust in AI-assisted science.

**Social Science:** Economist Sanchaita Hazra used AutoDiscovery to explore social and economic datasets, discovering that authors with doctoral degrees edited AI-generated abstracts substantially more than those with undergraduate or master's degrees—a significant finding her original research focus would have missed. She noted the tool condensed weeks of PhD-level regression analysis into hours.

## Availability and Impact

AutoDiscovery is now available as an experimental feature in AstaLabs. Researchers can sign up for Asta Preview to gain early access. The tool represents a paradigm shift in research methodology: rather than replacing expert intuition, it surfaces hidden questions in datasets, allowing researchers to focus effort where it matters most.

---

### Introducing AutoDiscovery: Automated scientific discovery, now in AstaLabs
- Published: 2026-02-12T08:00:00+00:00
- Products: 
- Tags: release, feature, platform, api
- URL: https://standarddb.com/ai2/allen-ai-releases-autodiscovery-automated-hypothesis-generation-tool-now--qbzqkzpx
- Original: https://allenai.org/blog/autodiscovery

## What Is AutoDiscovery?

AutoDiscovery transforms how scientists interact with data by inverting the traditional research workflow. Rather than starting with a hypothesis and testing it, researchers upload a structured dataset and let the system autonomously generate hypotheses in natural language, propose experiments, execute Python code, interpret statistical results, and use findings to generate new leads. The system has already produced publishable results across disciplines—from uncovering trophic relationships in marine ecosystem data to identifying mutual-exclusivity patterns in cancer mutations that could inform treatment decisions.

## How It Works

The system combines two key technical innovations:

- **Bayesian Surprise**: AutoDiscovery quantifies how much experimental evidence shifts its internal beliefs (represented as probability distributions). Rather than simply confirming expected findings, it prioritizes results that meaningfully contradict its prior expectations—reflecting the scientific intuition that surprising results often represent genuine discoveries.
- **Monte Carlo Tree Search (MCTS)**: The system uses intelligent search to navigate the vast space of possible scientific questions, balancing exploration of new hypotheses with deeper investigation of promising leads.

Together, these approaches guide AutoDiscovery toward unexpected patterns and non-obvious research directions that human researchers might overlook.

## Real-World Impact

Early partnerships demonstrate concrete value across domains. Researchers at the University of Utah used AutoDiscovery to accelerate exploratory analysis at unprecedented speed. At Fred Hutchinson Cancer Center, analyses that would normally require weeks or months were completed in a single day. A case study with oncologists at Swedish Cancer Institute's Paul G. Allen Research Center uncovered a mutual-exclusivity pattern between PIK3CA and TP53 mutations in breast cancer—a potential finding that could inform treatment strategies.

Several AutoDiscovery discoveries in social science were independently verified and published in a peer-reviewed paper (November 2025), demonstrating that the system's findings meet scientific publication standards.

## Availability and Next Steps

AutoDiscovery is now available as an experimental feature in AstaLabs within the Asta platform. Researchers can upload datasets and monitor discovery progress through an intuitive interface displaying hypotheses, surprisal scores, and statistical details. The system supports both quick analyses and extended overnight runs generating hundreds of experiments. Full case studies and technical documentation are available on Allen AI's website.

---

### v0.16.0rc3: [Bugfix] Fix MTP accuracy for GLM-5 (#34385)
- Published: 2026-02-12T04:54:27+00:00
- Products: 
- Tags: release, bugfix
- URL: https://standarddb.com/vllm/vllm-v0-16-0rc3-fixes-mtp-accuracy-issue-with-glm-5-model--vkcyjtvh
- Original: https://github.com/vllm-project/vllm/releases/tag/v0.16.0rc3

## vLLM v0.16.0rc3 Release

vLLM has released v0.16.0rc3, a release candidate version that includes a critical bugfix for Multi-Token Prediction (MTP) accuracy when using the GLM-5 model.

### What Changed

This release focuses on correcting an accuracy regression in the MTP implementation for GLM-5. The bugfix ensures that Multi-Token Prediction produces correct results when generating multiple tokens in parallel, which is essential for optimal performance on compatible hardware.

### Who This Affects

- Users running GLM-5 models on vLLM with MTP enabled
- Developers optimizing inference latency through parallel token generation
- Production deployments requiring high accuracy guarantees

### Next Steps

Users should update to v0.16.0rc3 or wait for the stable v0.16.0 release. If you're currently using GLM-5 with MTP features, verify accuracy on your workloads with this patched version.

---

### v0.5.88
- Published: 2026-02-12T00:02:45+00:00
- Products: Sim Studio
- Tags: release, feature, integration, bugfix, performance
- URL: https://standarddb.com/sim/sim-v0-5-88-adds-gemini-deep-research-integration-and-confluence-enhancements--wmmstrrj
- Original: https://github.com/simstudioai/sim/releases/tag/v0.5.88

## New Features

Sim Studio v0.5.88 introduces two significant capabilities:

**Gemini Deep Research Integration**: Developers can now leverage Google's Gemini Deep Research model directly through Sim's Interactions API. This enables workflows to access advanced research and analysis capabilities within automation pipelines.

**Confluence Enhancements**: The Confluence integration has been expanded with new operations including:
- List space labels
- Delete label functionality
- Delete page properties

## Performance & Stability Improvements

The release addresses infrastructure scaling with an increase to the trigger machine size, improving handling of larger workloads. Several UX bugs have been fixed, including cursor alignment issues in the variables block and unwanted tab-switching behavior from C, T, and E hotkeys.

## Bug Fixes

- Confluence integration now uses recommended query parameter patterns for improved reliability
- Variables block dropdown and cursor alignment corrected
- Tab-switching hotkeys (C, T, E) removed to prevent accidental navigation

---

### Cloudflare Fundamentals, Terraform - Terraform v5.17.0 now available
- Published: 2026-02-12T00:00:00+00:00
- Products: Cloudflare Workers, Cloudflare
- Tags: release, sdk, api, feature
- URL: https://standarddb.com/cloudflare/cloudflare-terraform-provider-v5-17-0-adds-ai-search-support-and-workers--ctsthhqk
- Original: https://developers.cloudflare.com/changelog/post/2026-02-12-terraform-v5170-provider

## Terraform v5.17.0 Release

Cloudflare has released Terraform v5.17.0 for its Terraform provider, continuing the rapid development pace established since the v5 launch in January 2025. The release includes new features, significant bug fixes, and foundational work for the v4-to-v5 migration tooling.

### New Features

This release introduces several new capabilities:

- **AI Search Integration**: New data sources for querying AI Search instances and tokens, bringing improved infrastructure-as-code support for Cloudflare's AI services
- **Workers Script Enhancements**: 
  - Support for `simple` rate limit binding
  - Targeted placement mode with `placement.target` array for specifying placement targets (region, hostname, host)
  - New `placement_mode` and `placement_status` computed fields for better control over script deployment
- **Authentication & Security**: Data sources for querying authenticated origin pull certificates, hostname-specific certificates, and related settings
- **Tenant Management**: Support for tenant unit management with new `unit` field in account resources
- **KV & Workers**: Enhanced data sources to retrieve KV values and script content directly

### Bug Fixes and Stability

The release addresses 25+ bug fixes focused on correctness and preventing state drift:

- Fixed field mapping issues across multiple resources (using correct `request_id`, `zone_id`, `account_id`, and `entry_id` fields instead of generic `id`)
- Improved handling of write-only fields and empty object normalization (particularly for Hyperdrive configs)
- Fixed authentication-related resources and lifecycle management
- Updated Zero Trust, pages, and DNS resources to match current API behavior

### Migration Path and Stability Roadmap

Cloudflare is laying groundwork for v4-to-v5 migration with state upgraders added to 95+ resources. Most-used resources are on track to reach stable status by end of March 2026, when a dedicated migration tool will be released to help users transition from v4 with ease.

---

### 1.3.3
- Published: 2026-02-12T00:00:00+00:00
- Products: 
- Tags: release, feature, api, bugfix
- URL: https://standarddb.com/modal/modal-sdk-1-3-3-adds-billing-reports-async-warnings-and-object-id-methods--gdyfbzhy
- Original: https://modal.com/docs/reference/changelog

## New Features

**Billing Reports**: Modal has promoted the `modal.billing.workspace_billing_report` API to General Availability and introduced a new `modal billing report` CLI command. This feature is available to all Team and Enterprise plan workspaces, enabling better cost tracking and visibility.

**Object ID Methods**: Two new convenience methods simplify referencing existing Modal objects:
- `modal.Queue.from_id()` - reference a Queue by its object id
- `modal.Dict.from_id()` - reference a Dict by its object id

## Breaking Changes & Warnings

**Async Usage Warnings Enabled by Default**: Modal's async context warnings are now enabled by default. These warnings fire when using blocking interfaces on Modal objects within async contexts—a pattern that can cause performance issues or bugs. The warnings include detailed, actionable suggestions for remediation.

To disable these warnings, set either:
- Environment variable: `MODAL_ASYNC_WARNINGS=0`
- Config file: `async_warnings = false` in `.modal.toml`

Modal recommends addressing warnings as they often indicate significant performance issues, but acknowledges they may occasionally be false positives. Users are encouraged to report any incorrect suggestions.

## Bug Fixes

**ASGI State Leak**: Fixed a bug where the ASGI scope's `state` contents could leak between requests when using `@modal.asgi_app`. This ensures proper isolation of request state in asynchronous web applications.

## Upgrade Notes

Review any async usage in your codebase for warnings and follow the provided remediation guidance. The warnings are verbose but intentionally designed to guide developers toward best practices.

---

### Cloudflare Fundamentals, Terraform - Terraform v5.17.0 now available
- Published: 2026-02-12T00:00:00+00:00
- Products: Cloudflare Workers, Cloudflare
- Tags: release, feature, sdk, platform
- URL: https://standarddb.com/cloudflare/cloudflare-terraform-provider-v5-17-0-adds-ai-search-support-and-targeted--fxtntccn
- Original: https://developers.cloudflare.com/changelog/2026-02-12-terraform-v5170-provider

## New Features and Capabilities

Cloudflare has released Terraform v5.17.0 with significant additions to provider functionality, particularly around AI Search integration and Workers script management. This release includes new data sources for AI Search instances and tokens, as well as enhanced capabilities for querying authenticated origin pull certificates and settings.

**Workers Script Enhancements**

The Workers script resource now supports targeted placement mode, allowing developers to specify placement targets using a new `placement.target` array for controlling placement by region, hostname, or host. Additional computed fields (`placement_mode` and `placement_status`) provide visibility into script placement. The provider also adds support for `simple` rate limit binding and the ability to retrieve script content directly via a new `script` field in the data source.

**Bug Fixes and Stability Improvements**

This release includes 20+ bug fixes addressing resource lifecycle issues, field normalization, and API compatibility. Notable fixes include correcting field references in `cloudforce_one_request_message`, DNS zone transfers, and DLP resources to use the proper identifiers instead of generic `id` fields. Additional improvements address state drift issues in hyperdrive configurations and empty object normalization.

**Migration and Stabilization Progress**

Cloudflare continues working toward complete resource stability, with the most-used resources targeted for stable status by end of March 2026. The provider now includes state upgraders for 95+ resources to support the ongoing migration from SDKv2 to the Framework, laying groundwork for improved v4-to-v5 migration tooling planned for the March release.

## What Developers Need to Do

Review the [full list of stabilized resources](https://github.com/cloudflare/terraform-provider-cloudflare/issues/6237) to identify which resources are production-ready. If using Workers scripts, update configurations to take advantage of targeted placement controls. Prepare for the March 2026 migration tool release by auditing your current v4 provider usage.

---

### Radar - Content Type Dimension for AI Bots in Cloudflare Radar
- Published: 2026-02-12T00:00:00+00:00
- Products: Cloudflare
- Tags: feature, api, platform
- URL: https://standarddb.com/cloudflare/cloudflare-radar-adds-content-type-insights-for-ai-bot-traffic-via-new-api--crrsyzmt
- Original: https://developers.cloudflare.com/changelog/2026-02-12-radar-ai-bots-content-type

## New Content Type Dimension for AI Bot Traffic

Cloudflare Radar now provides detailed insights into what types of content AI bots and crawlers are accessing across your properties. A new `content_type` dimension allows developers and security teams to understand the distribution of content served to AI crawlers, grouped by 16 MIME type categories.

## API Endpoints and Availability

The feature is accessible through two new endpoints:
- `/ai/bots/summary/content_type` - Get aggregated content type summaries
- `/ai/bots/timeseries_groups/content_type` - Track content type distribution over time

Beyond the API, individual bot information pages in the Radar dashboard now display content type distribution charts, enabling at-a-glance analysis of crawler behavior for both Verified Bots and AI Bots datasets.

## Content Type Categories

The dimension groups traffic into 16 categories including HTML, images, JSON, JavaScript, CSS, plain text, fonts, XML, YAML, video, audio, markdown, documents (PDF/Office), binary files (archives/executables), serialization formats (protobuf/gRPC), and a catch-all "Other" category. This granular classification helps identify what content types AI crawlers prioritize.

## Getting Started

Explore the data via the [AI Insights page](https://radar.cloudflare.com/ai-insights#content-type) in Cloudflare Radar or integrate the new API endpoints into your analytics workflows.

---

### February 12, 2026
- Published: 2026-02-12T00:00:00+00:00
- Products: 
- Tags: release, platform
- URL: https://standarddb.com/deepgram/deepgram-releases-self-hosted-february-2026-container-images-with-engine-v3-107--xdhnzwzf
- Original: https://developers.deepgram.com/changelog/2026/2/12

## Deepgram Self-Hosted February 2026 Release

Deepgram has published new container images for its self-hosted deployment option, tagged as release 260212. This release provides updated versions across all core components of the self-hosted infrastructure.

### Updated Components

The release includes:

- **Self-Hosted API**: version 1.177.3 (`quay.io/deepgram/self-hosted-api:release-260212`)
- **Self-Hosted Engine**: version 3.107.0-1 (`quay.io/deepgram/self-hosted-engine:release-260212`)
- **License Proxy**: version 1.9.2 (`quay.io/deepgram/self-hosted-license-proxy:release-260212`)
- **Billing**: version 1.12.1 (`quay.io/deepgram/self-hosted-billing:release-260212`)

### Important Requirements

Users deploying the engine component must ensure they have **NVIDIA driver version 570.172.08 or later** installed in their infrastructure to maintain compatibility.

### What Changed

This release includes general improvements to keep the Deepgram self-hosted software up-to-date. Users running self-hosted deployments should update their container images to the latest release for improved stability and performance.

---

### Cloudflare Fundamentals - Introducing Markdown for Agents
- Published: 2026-02-12T00:00:00+00:00
- Products: Cloudflare
- Tags: feature, api, platform
- URL: https://standarddb.com/cloudflare/cloudflare-launches-markdown-for-agents-auto-converts-html-via-content--vqtwswpk
- Original: https://developers.cloudflare.com/changelog/2026-02-12-markdown-for-agents

## Markdown for Agents Now Available

Cloudflare has introduced **Markdown for Agents**, a new platform feature that leverages HTTP content negotiation to serve web content in Markdown format to AI systems and agents. When an AI system makes a request with the `Accept: text/markdown` header, Cloudflare's network automatically converts HTML to Markdown on the fly at the source.

### How It Works

The implementation uses standard HTTP content negotiation:

- AI systems include `Accept: text/markdown` in their request headers
- Cloudflare's network detects this preference for enabled zones
- HTML content is converted to Markdown automatically and efficiently
- The response includes appropriate headers like `content-type: text/markdown` and metadata such as `x-markdown-tokens` for token counting

### Key Benefits

- **Efficiency**: Markdown's explicit structure reduces token waste compared to raw HTML
- **Performance**: Conversion happens at the network edge, not on client systems
- **Compatibility**: Uses standard HTTP negotiation, requiring no custom protocols
- **Transparent**: Works with any website on Cloudflare that enables the feature

### Getting Started

To use this feature, AI systems can make requests like:

```bash
curl https://example.com/page \
  -H "Accept: text/markdown"
```

Website operators can enable this feature for their zones and refer to the [Markdown for Agents documentation](https://developers.cloudflare.com/fundamentals/reference/markdown-for-agents/) for configuration details.

---

### Cloudflare WAN - Anycast IPs displayed on the dashboard
- Published: 2026-02-12T00:00:00+00:00
- Products: Cloudflare
- Tags: feature, platform
- URL: https://standarddb.com/cloudflare/cloudflare-wan-dashboard-now-displays-anycast-ips-for-tunnels--qmxdsnqm
- Original: https://developers.cloudflare.com/changelog/post/2026-02-12-anycast-ips-on-dashboard

## Improvement to Cloudflare WAN Dashboard

Cloudflare WAN has enhanced its tunnel configuration interface to display Anycast IP addresses directly in the dashboard. This update removes friction from the tunnel setup process for customers configuring IPsec and GRE connections.

### What Changed

The dashboard tunnel configuration form now **pre-loads Anycast IP addresses** automatically. Previously, customers had to retrieve these addresses through one of two methods:
- Receiving them during the onboarding process
- Making API calls to fetch them manually

### Benefits

This change addresses a real pain point in tunnel configuration:
- **Reduced setup friction**: Addresses are immediately visible without requiring API calls
- **Fewer configuration errors**: Pre-populated values minimize typos and mistakes
- **Improved UX**: Streamlined workflow for new tunnel deployments

### Action Required

None. This feature is automatically available to all Cloudflare WAN customers. Anycast IPs will appear in the tunnel configuration form without any additional steps.

For detailed guidance on configuring your tunnel endpoints, see the [Configure tunnel endpoints documentation](https://developers.cloudflare.com/cloudflare-wan/configuration/manually/how-to/configure-tunnel-endpoints/).

---

### Long-running Agents in Research Preview
- Published: 2026-02-12T00:00:00+00:00
- Products: 
- Tags: feature, platform, release
- URL: https://standarddb.com/cursor/cursor-launches-long-running-agents-to-handle-complex-multi-step-tasks--jjwjxdgy
- Original: https://cursor.com/changelog/02-12-26

## Long-Running Agents Now Available

Cursor has introduced long-running agents, an advancement that enables AI-powered development agents to tackle more complex tasks autonomously. Unlike standard agents that handle shorter, more limited tasks, these new long-running agents can work over extended horizons while maintaining context and making intelligent planning decisions.

## Key Capabilities

The long-running agents introduce several improvements:

- **Extended autonomy**: Work on larger, more complex tasks without requiring human intervention between steps
- **Strategic planning**: Agents plan their approach before executing, leading to better outcomes and more complete solutions
- **Improved PR quality**: Internal testing shows agents complete work previously considered too difficult, resulting in larger and more comprehensive pull requests with fewer follow-ups

## Availability and Access

Long-running agents are now available at [cursor.com/agents](https://cursor.com/agents) for:
- Ultra plan subscribers
- Teams plan subscribers
- Enterprise plan subscribers

The feature is currently in research preview, indicating Cursor is gathering feedback and may refine the experience based on real-world usage. Users can begin experimenting with the new agent capabilities immediately through their existing Cursor subscriptions.

---

### Introducing Theme Builder
- Published: 2026-02-12T00:00:00+00:00
- Products: 
- Tags: feature, platform, release
- URL: https://standarddb.com/zed/zed-launches-theme-builder-a-visual-editor-for-crafting-code-editor-themes--mgbkbchf
- Original: https://zed.dev/blog/theme-builder

## What's New

Zed has introduced **Theme Builder**, a visual editor that lets developers craft custom editor themes without touching a single line of JSON. Instead of manually tweaking hundreds of color tokens in configuration files, users can now use an interactive interface with a live preview to design their perfect theme.

## Key Features

The Theme Builder interface includes:

- **Live preview pane** showing an interactive replica of Zed's UI, updated instantly as you make changes
- **Inspector tool** that lets you right-click any element to see exactly which theme tokens control its appearance, eliminating guesswork
- **Color linking** that allows tokens to reference each other—change a source color and all linked tokens update automatically, with suggested links between related elements
- **Tree-sitter syntax highlighting** for pixel-perfect code previews that match Zed's actual highlighting
- **Import/export functionality** to load existing themes or generate exportable theme files

## The Problem Solved

Modern code editors like Zed now include 16 categories of UI colors, 10 categories of syntax colors, and player colors for multiplayer collaboration. Managing these interdependencies in raw JSON is tedious and error-prone. Color linking prevents visual inconsistencies, while the inspector eliminates the trial-and-error of determining which property controls which element.

## Getting Started

The Theme Builder includes several built-in themes (One, Ayu, Gruvbox) as starting points. Developers can import existing themes, customize them visually, and export the results either as [theme overrides](https://zed.dev/docs/themes#theme-overrides) or as extension files for sharing.

The tool is available at [zed.dev/theme-builder](https://zed.dev/theme-builder).

---

### Cloudflare Fundamentals - Introducing Markdown for Agents
- Published: 2026-02-12T00:00:00+00:00
- Products: Cloudflare
- Tags: feature, api, platform, integration
- URL: https://standarddb.com/cloudflare/cloudflare-launches-markdown-for-agents-enabling-ai-systems-to-request-html-to--lxjmjzmh
- Original: https://developers.cloudflare.com/changelog/post/2026-02-12-markdown-for-agents

## What's New

Cloudflare has introduced **Markdown for Agents**, a new feature that leverages HTTP content negotiation to serve markdown-formatted content to AI systems. When an AI agent requests a page from a Cloudflare-protected website with this feature enabled, it can express a preference for markdown format using the `Accept: text/markdown` header, and Cloudflare's edge network will automatically convert the HTML to markdown in real-time.

## How It Works

The implementation relies on standard HTTP content negotiation protocols. AI systems simply need to:

- Send an `Accept: text/markdown` header in their request
- Receive markdown-formatted content instead of HTML
- Benefit from optimized token usage and cleaner structured data

Example request:
```
curl https://example.com/page \
  -H "Accept: text/markdown"
```

The response includes a `content-type: text/markdown` header and additional metadata like `x-markdown-tokens` (showing token count) and content signals indicating the resource is suitable for AI training and input.

## Why This Matters

Markdown is becoming the standard interchange format for AI systems and agents due to its explicit structure and efficiency. By serving markdown directly from the edge, websites can:

- **Reduce token overhead** for AI processing
- **Improve output quality** from structured, cleaner formatting
- **Enable better content signal control** through `content-signal` headers (for training, search indexing, and AI input)

## Getting Started

Developers can enable Markdown for Agents on their Cloudflare zones and configure content signals to control how AI systems interact with their content. Full documentation is available in the [Cloudflare developer reference](https://developers.cloudflare.com/fundamentals/reference/markdown-for-agents/).

---

### Radar - Content Type Dimension for AI Bots in Cloudflare Radar
- Published: 2026-02-12T00:00:00+00:00
- Products: Cloudflare
- Tags: feature, api, platform
- URL: https://standarddb.com/cloudflare/cloudflare-radar-adds-content-type-dimension-for-ai-bot-traffic-analysis--dxscrjxf
- Original: https://developers.cloudflare.com/changelog/post/2026-02-12-radar-ai-bots-content-type

## New Content Type Dimension for AI Bots

Cloudflare Radar has introduced a new `content_type` dimension and filter that provides detailed insights into the types of content being accessed by AI bot and crawler traffic. This addition helps developers and site operators understand the composition of requests from AI crawlers across their infrastructure.

## API Endpoints and Data Access

The feature is available through two new API endpoints:

- **`/ai/bots/summary/content_type`** - Returns aggregated content type distribution for AI bot traffic
- **`/ai/bots/timeseries_groups/content_type`** - Provides time-series data grouped by content type

These endpoints allow developers to programmatically retrieve detailed analytics about what content types AI crawlers are consuming, enabling more granular traffic analysis and optimization.

## Content Type Categories

The data is organized into 16 distinct content type categories, including:

- **Web content**: HTML, Images, JavaScript, CSS, Fonts
- **Data formats**: JSON, XML, YAML, Plain Text, Markdown
- **Media**: Video, Audio
- **Documents**: PDFs, Office documents, ePub, CSV
- **Technical formats**: Binary executables, archives, WebAssembly, Protobuf, gRPC, MessagePack
- **Other**: Catch-all for remaining content types

## Dashboard Integration

Beyond the API, individual bot information pages in the Cloudflare Radar interface now display content type distribution charts for verified AI crawlers. The AI Insights page includes a dedicated section to explore this data, making it easy for users to understand crawler behavior without API calls.

---

### Security Center - Enhanced Logo Matching for Brand Protection
- Published: 2026-02-12T00:00:00+00:00
- Products: Cloudflare
- Tags: feature, security
- URL: https://standarddb.com/cloudflare/cloudflare-enhances-logo-matching-with-configurable-thresholds-starting-at-75--jkjlxpxh
- Original: https://developers.cloudflare.com/changelog/post/2026-02-12-brand-protection-logo-matching-percentage-selector

## Enhanced Logo Matching Capabilities

Cloudflare has significantly upgraded the Logo Matching functionality within Brand Protection. Previously limited to detecting near-perfect (approximately 100%) matches, the platform now leverages a redesigned matching model and user interface to identify a wider range of brand asset variations and sophisticated impersonations.

### New Features

**Configurable Match Thresholds**: Users can now set custom minimum match scores starting at 75% when creating logo queries. This flexibility enables detection of subtle variations and high-quality impersonations that would previously go undetected.

**Visual Match Scores**: The platform now displays exact match percentages directly in results tables, using color-coded lozenges to indicate threat severity at a glance.

**Direct Logo Previews**: Logo previews are now available directly in the Cloudflare dashboard, matching the functionality of string matching tools and allowing for quick verification of infringements without external tools.

### Benefits for Security Teams

These enhancements enable organizations to **expose sophisticated threat actors** who use slightly altered logos to bypass basic detection filters. The visual indicators and configurable thresholds significantly **reduce triage time**, allowing security teams to prioritize the most relevant threats immediately rather than manually reviewing every match.

For more details, refer to the [Brand Protection documentation](https://developers.cloudflare.com/security-center/brand-protection/).

---

### Security Center - Enhanced Logo Matching for Brand Protection
- Published: 2026-02-12T00:00:00+00:00
- Products: Cloudflare
- Tags: feature, security
- URL: https://standarddb.com/cloudflare/cloudflare-enhances-logo-matching-for-brand-protection-with-configurable--fsqmnlsp
- Original: https://developers.cloudflare.com/changelog/2026-02-12-brand-protection-logo-matching-percentage-selector

## Enhanced Logo Matching for Brand Protection

Cloudflare has significantly upgraded its Logo Matching capabilities within the Security Center's Brand Protection feature. The system previously detected only near-perfect (100%) matches, but now uses a redesigned matching model that captures a much wider range of brand asset variations and sophisticated impersonations.

### What's New

**Configurable match thresholds** allow security teams to set a minimum match score starting at 75% when creating logo queries. This flexibility enables detection of both subtle variations and high-quality impersonations that would previously slip through detection.

**Visual match scores** are now displayed directly in the results table as color-coded lozenges, indicating the exact percentage match and severity level at a glance. **Direct logo previews** in the Cloudflare dashboard provide quick visual verification of potential infringements, similar to string match functionality.

### Key Benefits

The enhanced system helps security teams expose sophisticated impersonators who alter logos to bypass basic detection filters. The visual severity indicators and color-coded results significantly reduce triage time, allowing teams to prioritize the most relevant threats immediately rather than manually reviewing all matches.

For more details, refer to the [Brand Protection documentation](https://developers.cloudflare.com/security-center/brand-protection/).

---

### GLM-5 is live on AI Gateway
- Published: 2026-02-11T13:00:00+00:00
- Products: Vercel AI SDK, Vercel
- Tags: feature, model, integration, api
- URL: https://standarddb.com/vercel/vercel-ai-gateway-now-supports-glm-5-model-with-native-agentic-capabilities--jxldbmqp
- Original: https://vercel.com/changelog/glm-5-is-live-on-ai-gateway

## GLM-5 Now Available on AI Gateway

Vercel has added GLM-5 from Z.AI to its AI Gateway platform, making the model directly accessible through Vercel's unified API without requiring separate provider authentication.

### Key Capabilities

GLM-5 builds on its predecessor GLM-4.7 with several notable improvements:

- **Multiple thinking modes** for complex reasoning tasks
- **Enhanced long-range planning and memory** for maintaining context across extended interactions
- **Improved multi-step agent task handling** designed for autonomous workflows
- **Strong agentic capabilities** including autonomous tool use and structured reasoning
- **Document analysis** optimized for extracting data from contracts, financial reports, and similar documents
- **Agentic coding** with comprehensive API generation including authentication and test coverage

### Getting Started

Developers can immediately use GLM-5 through the Vercel AI SDK by setting the model identifier to `zai/glm-5`. The integration works seamlessly with existing AI Gateway features including usage tracking, cost monitoring, automatic retries, failover configuration, and performance optimizations.

### AI Gateway Benefits

By accessing GLM-5 through AI Gateway, developers gain additional infrastructure value: unified API for multiple model providers, built-in observability, Bring Your Own Key support, intelligent routing with automatic retries, and higher uptime guarantees compared to direct provider access.

---

### Advanced egress firewall filtering for Vercel Sandbox
- Published: 2026-02-11T13:00:00+00:00
- Products: Vercel
- Tags: feature, api, security, platform
- URL: https://standarddb.com/vercel/vercel-sandbox-adds-sni-based-egress-firewall-for-controlling-outbound-network--qvrylyzt
- Original: https://vercel.com/changelog/advanced-egress-firewall-filtering-for-vercel-sandbox

## Host-Based Network Egress Control

Vercel Sandbox now enforces egress network policies through **SNI-peeking firewalls** and CIDR-based rules, giving developers precise control over which external services sandboxed code can reach. Unlike traditional IP-based firewall rules that cannot distinguish between thousands of domains sharing a handful of IP addresses, this SNI approach inspects the initial bytes of TLS handshakes to extract target hostnames, making it effective across modern encrypted protocols like HTTPS, Redis, and PostgreSQL.

## Configuration and Workflow

Developers define allowlisted domains at sandbox creation time or update policies dynamically on running instances without restarts:

```typescript
const sandbox = await Sandbox.create({
  networkPolicy: {
    allow: ['ai-gateway.vercel.sh', '*.vercel.com'],
  },
});
```

This enables flexible multi-phase workflows: open the network for dependency installation, lock it down to specific services during untrusted code execution, selectively open egress for results, and fully air-gap with `deny-all` when complete.

## Security and Availability

Outbound TLS connections are matched against policies at the handshake phase, with unauthorized destinations rejected before any data transmission. By default, sandboxes have unrestricted internet access; developers can now enforce deny-by-default posture to prevent compromised or hallucinated code snippets from exfiltrating data or making unintended API calls.

The feature is available on all Vercel plans. Wildcard support simplifies configuration for CDN-backed services, and legacy non-TLS systems fall back to IP/CIDR rules. Full documentation covers network egress firewall policy configuration and use cases.

---

### Vercel Flags is now in public beta
- Published: 2026-02-11T13:00:00+00:00
- Products: Next.js, Vercel
- Tags: feature, platform, sdk, pricing
- URL: https://standarddb.com/vercel/vercel-flags-moves-to-public-beta-with-native-next-js-and-sveltekit-support--kbnsgkpg
- Original: https://vercel.com/changelog/vercel-flags-is-now-in-public-beta

## Overview

**Vercel Flags** is now in public beta, bringing native feature flag management directly to the Vercel platform. This service allows teams to create, manage, and deploy feature flags with targeting rules and user segmentation without relying on external services.

## Key Features

The Flags SDK provides framework-native integration for Next.js and SvelteKit applications:

- **Simple API**: Define flags in TypeScript with the `flag()` function and use them directly in your application code
- **Dashboard Management**: Create and manage feature flags, targeting rules, and user segments directly in the Vercel Dashboard
- **Environment Controls**: Apply different flag configurations across development, preview, and production environments
- **OpenFeature Support**: Teams using other frameworks or custom backends can integrate via the OpenFeature standard, enabling consistent flag management across multiple systems

## Implementation & Pricing

The integration is straightforward—define flags in a `flags.ts` file using the Vercel adapter, then reference them in your pages as simple async functions. For non-Next.js/SvelteKit applications, the OpenFeature provider allows seamless integration with existing feature flag workflows.

Pricing is set at **$30 per million flag requests** ($0.00003 per event), where a flag request represents any application request that reads the flags configuration. Multiple flags evaluated in a single request count as one flag request.

## Availability

Vercel Flags is now available in public beta to all teams on any Vercel plan. Developers can get started immediately by creating a Vercel account or upgrading their existing deployment.

---

### v0.16.0rc2: Patch protobuf for CVE-2026-0994 (#34253)
- Published: 2026-02-11T10:33:40+00:00
- Products: 
- Tags: 
- URL: https://standarddb.com/vllm/v0-16-0rc2-patch-protobuf-for-cve-2026-0994-34253--mdznvcjx
- Original: https://github.com/vllm-project/vllm/releases/tag/v0.16.0rc2



---

### MolmoSpaces, an open ecosystem for embodied AI
- Published: 2026-02-11T08:00:00+00:00
- Products: 
- Tags: release, open-source, platform, feature
- URL: https://standarddb.com/ai2/ai2-launches-molmospaces-an-open-simulation-platform-for-embodied-ai-with-230--zghcgqzy
- Original: https://allenai.org/blog/molmospaces

## A Unified Platform for Embodied AI Research

AI2 has launched **MolmoSpaces**, a comprehensive open ecosystem designed to accelerate embodied AI research by providing the data, assets, and benchmarking infrastructure that the field has lacked. The platform unifies over 230,000 indoor scenes and more than 130,000 object models—curated from Objaverse and AI2's THOR environment—with over 42 million annotated robotic grasps. All assets are provided in standard formats (MJCF and USD) and are compatible with popular simulators including MuJoCo, ManiSkill, and NVIDIA Isaac Lab/Sim.

## High-Fidelity Physics and Realistic Simulation

A key differentiator of MolmoSpaces is its commitment to physics accuracy. Unlike earlier environments that used simplified physics and "magic grasps," MolmoSpaces leverages physics engines with carefully validated physical parameters. For rigid objects, mass and density are verified through LLM-assisted annotation and adjusted iteratively. Articulated objects undergo teleoperation-based tuning to ensure realistic joint behavior, with the simulation itself validated against real robotic trajectories. All colliders and mesh preparation are manually annotated for stable contact-rich simulation, using convex decomposition for high-fidelity contact where needed.

## Systematic Benchmarking for Generalization

MolmoSpaces includes **MolmoSpaces-Bench**, a comprehensive benchmark that enables researchers to measure policy generalization systematically across multiple axes: object properties (shape, size, weight, articulation), layouts (multi-room, multi-floor, clutter), task complexity, sensory conditions (lighting, viewpoints), dynamics, and task semantics. This controlled-variation approach lets researchers probe robustness to specific factors—such as grasp performance across object masses or policy resilience to lighting changes—rather than reporting a single aggregate success rate. The benchmark also supports sim-to-real validation through real-world testing.

## Scale and Modularity

The platform's assets span both hand-curated THOR objects and carefully filtered Objaverse assets (129,000 objects across ~3,000 WordNet synsets). Scenes come from multiple sources—iTHOR-120, ProcTHOR-10K, ProcTHOR-Objaverse, and Holodeck—representing homes, offices, classrooms, hospitals, and museums. All 42+ million grasps are 6-DoF poses sampled directly from realistic object geometry, tested for diversity and robustness through perturbation and actuation analysis. An accompanying trajectory-generation pipeline enables reproducible demonstrations at scale.

## Getting Started

MolmoSpaces is fully open-source and modular, with code on GitHub, datasets on Hugging Face, a technical report, and a live demo. Researchers can inspect and modify MJCF files, regenerate grasps, swap robots and controllers, and use the platform across multiple simulators—providing the foundation for large-scale embodied AI research.

---

### v0.5.87
- Published: 2026-02-11T06:47:11+00:00
- Products: Sim Studio, Sim
- Tags: release, bugfix
- URL: https://standarddb.com/sim/sim-releases-v0-5-87-with-authentication-workflow-fix--rbkzgjtp
- Original: https://github.com/simstudioai/sim/releases/tag/v0.5.87

## Release Overview

Sim v0.5.87 is now available with a targeted bug fix for the authentication workflow system.

## What Changed

- **Authentication Fix**: Resolved an issue in the workflow system handler for authentication processes

## Action Items

Users experiencing issues with authentication workflows should update to v0.5.87 to receive this fix.

---

### v0.5.86
- Published: 2026-02-11T06:05:51+00:00
- Products: Sim Studio
- Tags: release, feature, api, integration
- URL: https://standarddb.com/sim/sim-releases-v0-5-86-with-copilot-mcp-server-side-execution-and-enterprise--gwlhtjjx
- Original: https://github.com/simstudioai/sim/releases/tag/v0.5.86

## Key Features

**Copilot Enhancements**: This release brings significant improvements to the copilot functionality with two major additions:
- **MCP Integration**: Copilot now supports Model Context Protocol (MCP) with server-side execution, enabling more robust and scalable copilot operations
- **Enterprise Configuration**: New enterprise-level configuration options for copilot deployments, including support for copilot-only Helm deployments

## Improvements & Observability

The release focuses on developer experience and operational visibility:

- **Logging Enhancements**: Skill icons are now displayed in trace spans for better visual identification within logs
- **Workflow Logs**: Increased the workflow logs limit from 1,000 to 5,000 entries per workflow, providing more comprehensive debugging information
- **Integrations**: Added 1Password integration and improved JSM/Jira integration with destructured outputs
- **MCP Stability**: Enhanced MCP SSE (Server-Sent Events) notifications with improved file handling in Jira
- **Preview Deployments**: Added trigger mode context for deploy previews

## Bug Fixes & Stability

Several critical fixes ensure production reliability:
- Fixed execution state scoping to prevent cross-workflow data leakage
- Hardened the notification system against race conditions
- Improved Slack file metadata resolution with partial event payloads
- Fixed subflow logs rendering in the terminal UI
- Upgraded Bun runtime from 1.3.3 to 1.3.9

---

### Workers - Workers are no longer limited to 1000 subrequests
- Published: 2026-02-11T00:00:00+00:00
- Products: Cloudflare Workers
- Tags: feature, api, platform
- URL: https://standarddb.com/cloudflare/cloudflare-workers-lifts-1000-subrequest-limit-to-10-000-by-default--zbmqbjgy
- Original: https://developers.cloudflare.com/changelog/2026-02-11-subrequests-limit

## Higher Limits for Production Workloads

Cloudflare Workers has removed the longstanding 1000-subrequest limit per invocation, a constraint that frequently blocked complex applications and long-running tasks. This change directly addresses pain points for developers running open websockets on Durable Objects or extended Workflows that naturally exceed the previous ceiling.

## New Default and Configuration Options

- **Paid plans** now default to **10,000 subrequests** per invocation
- Limits can be increased up to **10 million** via Wrangler configuration
- **Free plan** users remain constrained: 50 external subrequests + 1,000 Cloudflare service requests per invocation
- Both `wrangler.jsonc` (JSON) and `wrangler.toml` (TOML) configuration formats are supported

## Configuration and Safety Controls

You can set subrequest limits in your Wrangler configuration file using either format:

```toml
[limits]
subrequests = 50_000
```

To protect against runaway code or unexpected costs, developers can also enforce stricter limits:

```toml
[limits]
subrequests = 10
cpu_ms = 1_000
```

This allows fine-grained control over resource consumption across different environments or use cases. Refer to the [Wrangler configuration documentation](https://developers.cloudflare.com/workers/wrangler/configuration/#limits) and [subrequest limits guide](https://developers.cloudflare.com/workers/platform/limits/#subrequests) for additional details.

---

### Apps can now display thinking steps to users!
- Published: 2026-02-11T00:00:00+00:00
- Products: 
- Tags: feature, api, sdk
- URL: https://standarddb.com/slack/slack-introduces-task-cards-and-plan-blocks-for-displaying-ai-app-workflows--cgvnxpts
- Original: https://docs.slack.dev/changelog/2026/02/11/task-cards-plan-blocks

## New Block Kit Components

Slack has introduced three new components to help developers showcase app and agent workflows:

- **task_card block**: Displays individual tasks with clear visual hierarchy
- **url_source element**: Embeds clickable URL references directly within task cards
- **plan block**: Aggregates multiple task cards into a unified, comprehensive view

## Streaming API Enhancements

The text-streaming API methods now support two powerful new parameters:

- **task_display_mode**: Controls whether tasks are presented individually or as part of a complete plan view
- **chunks**: Allows you to update task cards and plans dynamically, with support for both structured task updates and plain markdown content

## Getting Started

Developers can begin implementing these features using the existing text-streaming methods:
- `chat.startStream`
- `chat.appendStream`
- `chat.stopStream`

The Slack documentation includes a dedicated guide on [developing AI apps](https://docs.slack.dev/ai/developing-ai-apps#streaming) with working examples and best practices for leveraging these new components.

---

### February 11, 2026
- Published: 2026-02-11T00:00:00+00:00
- Products: 
- Tags: platform, feature, api
- URL: https://standarddb.com/deepgram/deepgram-triples-default-concurrency-limits-for-streaming-speech-and-text-apis--tlsfzqbc
- Original: https://developers.deepgram.com/changelog/2026/2/11

## Increased Concurrency Limits

Deepgram is raising default concurrency limits across three core services:

- **Streaming Speech to Text**
- **Text to Speech**
- **Voice Agent**

### What's Changing

Default concurrency limits are being increased by up to **3X** for customers on the following plans:
- Pay as you Go
- Growth
- Enterprise

### What You Need to Know

This change enables higher throughput for production workloads without requiring plan upgrades or manual limit adjustments. Developers using these services should see improved performance and capacity for concurrent requests.

For specific rate limit details applicable to your plan, consult the [API Rate Limits documentation](https://developers.deepgram.com/reference/api-rate-limits).

### Action Items

- Review your current concurrency usage to take advantage of the increased limits
- No code changes are required; limits are updated automatically on your account

---

### Workers - Workers are no longer limited to 1000 subrequests
- Published: 2026-02-11T00:00:00+00:00
- Products: Cloudflare Workers
- Tags: feature, platform, api
- URL: https://standarddb.com/cloudflare/cloudflare-workers-lifts-1-000-subrequest-limit-paid-plans-now-support-up-to-10--qnjllljq
- Original: https://developers.cloudflare.com/changelog/post/2026-02-11-subrequests-limit

## Subrequest Limit Increased Significantly

Cloudflare Workers is no longer constrained by the previous 1,000 subrequest limit per invocation. This change enables developers to make substantially more `fetch()` calls and requests to Cloudflare services within a single Worker request without hitting hard limits.

## New Default Limits and Configuration

**Paid plan defaults:** Workers on paid plans now have a default limit of 10,000 subrequests per invocation, with the ability to increase this up to 10 million through Wrangler configuration. **Free plan:** Free tier users remain limited to 50 external subrequests and 1,000 subrequests to Cloudflare services per invocation.

Developers can configure custom subrequest limits in their `wrangler.jsonc` or `wrangler.toml` file:

```json
{
  "limits": {
    "subrequests": 50000
  }
}
```

## Cost and Safety Controls

To prevent runaway code or unexpected billing, developers can also set conservative limits lower than the defaults for both subrequests and CPU milliseconds:

```json
{
  "limits": {
    "subrequests": 10,
    "cpu_ms": 1000
  }
}
```

This update directly addresses the needs of **long-running Workers workloads** such as open WebSocket connections on Durable Objects and long-running Workflows, which frequently exceeded the previous ceiling. Refer to the [Wrangler configuration documentation](https://developers.cloudflare.com/workers/wrangler/configuration/#limits) and [subrequest limits guide](https://developers.cloudflare.com/workers/platform/limits/#subrequests) for complete details.



---

### GitHub Mobile: Model picker for Copilot coding agent
- Published: 2026-02-11T00:00:00+00:00
- Products: GitHub Copilot, GitHub Dependabot
- Tags: feature, release, model
- URL: https://standarddb.com/github/github-mobile-gains-copilot-model-picker-for-on-the-go-coding-sessions--qbfmltjy
- Original: https://github.blog/changelog/2026-02-11-github-mobile-model-picker-for-copilot-coding-agent

## GitHub Mobile Model Picker Now Available

GitHub has rolled out a model picker feature for the Copilot coding agent in its mobile app, enabling subscribers to select which AI model powers their coding sessions on-the-go.

### Available Models

Users can choose from:

- **Auto** (optimizes for speed and performance based on availability)
- **Anthropic Claude Opus 4.5** and **Claude Opus 4.6**
- **Anthropic Claude Sonnet 4.5**
- **OpenAI GPT-5.1-Codex-Max** and **GPT-5.2-Codex**

### Current Availability and Rollout

The feature is now generally available for **Copilot Pro and Pro+ subscribers** on both iOS and Android platforms. Support for Copilot Business and Enterprise customers is coming soon.

### Key Benefits

This addition addresses a common mobile workflow gap: developers can now match model selection to their immediate needs without returning to desktop. Users seeking faster responses for quick tasks can select a lightweight model, while deeper work can leverage more capable alternatives like Claude Opus 4.6.

Developers should update to the latest production build of GitHub Mobile to access this feature.

---

### Workers - Improved React Server Components support in the Cloudflare Vite plugin
- Published: 2026-02-11T00:00:00+00:00
- Products: Cloudflare Workers
- Tags: feature, integration, sdk
- URL: https://standarddb.com/cloudflare/cloudflare-vite-plugin-adds-seamless-react-server-components-support-via--lbpmgsdv
- Original: https://developers.cloudflare.com/changelog/post/2026-02-11-vite-plugin-child-environments

## React Server Components Support

The Cloudflare Vite plugin now provides seamless integration with **@vitejs/plugin-rsc**, the official Vite plugin for React Server Components. This integration enables developers to build modern RSC applications directly on Cloudflare Workers.

## New childEnvironments Configuration Option

A `childEnvironments` option has been added to the plugin configuration, allowing developers to use multiple environments within a single Worker. This enables the parent environment to import modules from child environments, creating separate module graphs for RSC use cases.

**Example configuration:**
```javascript
export default defineConfig({
  plugins: [
    cloudflare({
      viteEnvironment: {
        name: "rsc",
        childEnvironments: ["ssr"],
      },
    }),
  ],
});
```

## Framework Integration

The underlying functionality provided by @vitejs/plugin-rsc powers frameworks like React Router that build upon it for RSC support. Cloudflare has included a [basic example in the plugin repository](https://github.com/vitejs/vite-plugin-react/tree/main/packages/plugin-rsc/examples/starter-cf-single) to help developers get started with RSC on Cloudflare Workers.

This enhancement makes it easier for developers to leverage React Server Components' benefits—including improved performance, better security, and reduced client-side JavaScript—while deploying to Cloudflare's global edge network.

---

### Workers - Improved React Server Components support in the Cloudflare Vite plugin
- Published: 2026-02-11T00:00:00+00:00
- Products: Cloudflare Workers
- Tags: feature, sdk, integration
- URL: https://standarddb.com/cloudflare/cloudflare-vite-plugin-adds-react-server-components-support-with-child--cbgkntbj
- Original: https://developers.cloudflare.com/changelog/2026-02-11-vite-plugin-child-environments

## React Server Components Integration

The Cloudflare Vite plugin now provides seamless integration with `@vitejs/plugin-rsc`, the official Vite plugin for React Server Components. This enables developers to build RSC-based applications on Cloudflare Workers with improved tooling support.

## New Child Environments Feature

A `childEnvironments` option has been added to the plugin configuration, allowing developers to use multiple environments within a single Worker. This enables the parent environment to import modules from child environments, creating separate module graphs as needed.

### Configuration Example

```javascript
export default defineConfig({
  plugins: [
    cloudflare({
      viteEnvironment: {
        name: "rsc",
        childEnvironments: ["ssr"],
      },
    }),
  ],
});
```

## Framework Support

The `@vitejs/plugin-rsc` provides foundational functionality that modern frameworks like React Router build upon. Developers can reference the official [Cloudflare example](https://github.com/vitejs/vite-plugin-react/tree/f066114c3e6bf18f5209ff3d3ef6bf1ab46d3866/packages/plugin-rsc/examples/starter-cf-single) in the Vite plugin repository to get started with RSC-powered applications on Workers.

---

### add-mcp: Install MCP Servers Across Coding Agents and Editors
- Published: 2026-02-10T19:47:14+00:00
- Products: 
- Tags: feature, sdk, integration, open-source, platform
- URL: https://standarddb.com/neon/neon-releases-add-mcp-cli-to-install-mcp-servers-across-coding-agents-and--bptvqgds
- Original: https://neon.com/blog/add-mcp

## What is add-mcp?

**add-mcp** is a command-line tool that streamlines MCP (Model Context Protocol) server installation across different coding agents and editors. Inspired by Vercel's `add-skill` tool, it solves a real pain point: each editor and coding agent stores MCP configurations differently, requiring developers to manually configure the same server multiple times.

## How It Works

The tool operates with a single command syntax. To install an MCP server, users simply run:

```
npx add-mcp https://mcp.context7.com/mcp
```

or for npm-published servers:

```
npx add-mcp next-devtools-mcp@latest
```

The CLI automatically:
- Detects which coding agents are installed on your system
- Identifies which agents are configured in your project
- Writes correct configuration files for each tool
- Supports both project-level and global installations

## Supported Environments

add-mcp v1 supports installation across:
- Claude Code and Claude Desktop
- Cursor
- VS Code
- Codex
- Gemini CLI
- Goose
- OpenCode
- Zed

Users can explicitly target specific agents with the `-a` flag or use `-y` to skip all interactive prompts.

## Key Features

**Flexible configuration options:**
- Project-level or global installation (`-g` flag)
- Selective agent targeting (`-a cursor -a claude-code`)
- Non-interactive mode for automation (`-y` flag)

**For MCP maintainers:** add-mcp is designed as a shared ecosystem building block, reducing documentation burden and standardizing installation experiences across the growing MCP ecosystem.

---

### Agents can now access runtime logs with Vercel's MCP server
- Published: 2026-02-10T13:00:00+00:00
- Products: Vercel
- Tags: feature, api, integration
- URL: https://standarddb.com/vercel/vercel-s-mcp-server-adds-runtime-log-access-for-ai-agents--wfymrgxq
- Original: https://vercel.com/changelog/agents-can-now-access-runtime-logs-with-vercels-mcp-server

## New Runtime Log Access via MCP Server

Vercel has introduced the **get_runtime_logs** tool within its MCP server, enabling AI agents to directly access runtime logs for any project or deployment. This integration allows agents to programmatically retrieve logs generated by Vercel Functions invocations across both preview and production environments.

## Use Cases for Agents

The new capability unlocks several debugging and monitoring workflows for agents:

- **Debug failing requests** — Agents can automatically investigate production issues by retrieving relevant logs
- **Inspect function output** — Access console.log messages and function execution traces
- **Search and filter logs** — Query logs for specific errors, request IDs, or patterns
- **Cross-deployment analysis** — Compare behavior across multiple deployments to identify regressions

## Getting Started

Developers can start using this feature immediately through the [Vercel MCP server documentation](https://vercel.com/docs/mcp/vercel-mcp). The tool integrates seamlessly with the existing MCP protocol, making it easy to incorporate into agent-based workflows and automation pipelines.

This enhancement positions Vercel Functions as a more observable and agent-friendly platform, particularly for teams building AI-powered debugging and monitoring systems.

---

### Support for Sign in with Apple
- Published: 2026-02-10T13:00:00+00:00
- Products: Vercel
- Tags: feature, platform
- URL: https://standarddb.com/vercel/vercel-adds-sign-in-with-apple-support--pndgnrps
- Original: https://vercel.com/changelog/sign-in-with-apple-support

## Sign in with Apple Now Available

Vercel has added support for Sign in with Apple to its authentication system, streamlining login for users with Apple accounts.

### How It Works

**Automatic Account Linking**: If your Apple account uses an Apple email domain (@icloud.com, @mac.com, @me.com, etc.) that matches your existing Vercel account email, simply click the Apple button on the login screen to authenticate. Your accounts will be automatically linked.

**Manual Connection**: If your Apple email doesn't match your Vercel account email, you can manually connect your Apple account from your account settings once logged in.

### Getting Started

Visit the login screen to use Sign in with Apple, or manage your authentication methods in your [account settings](https://vercel.com/account/settings/authentication).

---

### PostHog joins the Vercel Marketplace
- Published: 2026-02-10T13:00:00+00:00
- Products: Next.js, Vercel
- Tags: integration, feature, release
- URL: https://standarddb.com/vercel/posthog-joins-vercel-marketplace-with-feature-flags-and-a-b-testing-integration--ssmpnwww
- Original: https://vercel.com/changelog/posthog-joins-the-vercel-marketplace

## PostHog Integration Now Available

PostHog is now an official provider in the Vercel Marketplace, bringing advanced feature flagging and experimentation capabilities to developers building on Vercel. This integration connects PostHog's feature flags, experimentation, and analytics platform directly into your development workflow.

## Key Capabilities

With the PostHog integration, you can:

- **Declare flags in code** using the Flags SDK and the dedicated `@flags-sdk/posthog` adapter
- **Toggle features in real time** for specific users or cohorts without redeployment
- **Gradual rollouts** using percentage-based feature releases to minimize risk
- **Run A/B tests** to validate impact before full production releases

## Getting Started

The integration provides a straightforward developer experience. Create a `flags.ts` file with an identify function to reference user context, then define flags with the PostHog adapter. The Flags SDK handles the abstraction layer, making feature flag management consistent across different providers.

Vercel has published a [PostHog template](https://vercel.com/templates/edge-middleware/posthog-with-flags-sdk-and-next-js) to help you get started quickly with example implementations and best practices.

## Use Cases

This integration is designed to help teams **test in production, reduce release risk, and make data-driven decisions** based on real user behavior—all within your existing Vercel deployment platform. It's particularly useful for teams practicing progressive deployments and validation before full rollouts.

---

### vercel logs CLI command now optimized for agents with historical log querying
- Published: 2026-02-10T13:00:00+00:00
- Products: Vercel
- Tags: feature, api, platform
- URL: https://standarddb.com/vercel/vercel-s-logs-cli-command-gains-historical-querying-and-agent-friendly-filters--glnddflf
- Original: https://vercel.com/changelog/vercel-logs-cli-command-now-optimized-for-agents-with-historical-log

## Enhanced Log Querying for Developers and Agents

Vercel has significantly upgraded the `vercel logs` CLI command with new querying and filtering capabilities. The rebuilt command makes it easier to search and retrieve logs across your projects, with built-in support for agent workflows that need programmatic access to historical log data.

## New Filtering Capabilities

The enhanced command supports multiple filtering options to help you find exactly what you need:

- **Project-level filtering** to scope logs to specific projects
- **Deployment ID filtering** to trace logs from particular deployments
- **Request ID filtering** for detailed request tracking
- **Arbitrary string search** for flexible pattern matching across logs

## Smart Context Detection

A key improvement is automatic git context detection. When you run `vercel logs` from a project directory, the command automatically scopes results to your current repository—eliminating the need to manually specify project details and streamlining local debugging workflows.

## Use Cases

This update is particularly valuable for:

- **Debugging production issues** with precise log retrieval
- **Building automated monitoring workflows** through the CLI
- **Agent-driven monitoring** where AI tools need programmatic log access with filtering capabilities

You can explore the full capabilities in the [Vercel CLI documentation for the logs command](https://vercel.com/docs/cli/logs).

---

### MCP server auto-discovery
- Published: 2026-02-10T09:00:00+00:00
- Products: 
- Tags: feature, api, integration, platform
- URL: https://standarddb.com/replicate/replicate-s-mcp-server-now-auto-discoverable-via-official-registry--gqbmycdk
- Original: https://replicate.com/changelog/2026-02-10-mcp-server-auto-discovery

## Auto-Discovery Implementation

Replicate's MCP server can now be discovered automatically through the official [MCP Registry](https://modelcontextprotocol.io/registry/about), an official metadata repository backed by Anthropic, GitHub, and Microsoft. The implementation follows standard specifications by exposing a `/.well-known/mcp/server.json` endpoint that publishes comprehensive metadata about the MCP server.

## How It Works

The MCP Registry doesn't host code directly—it serves as a centralized metadata repository that describes where to find servers and how to install them. When you publish a `server.json` file at the well-known endpoint, the Registry can discover your server automatically, and MCP clients use the Registry to find and install servers without manual configuration.

## Client Integration

Several MCP clients now have built-in discovery capabilities:

- **VS Code**: Best Registry integration available. Enable `chat.mcp.gallery.enabled` in settings, then search `@mcp` in the Extensions view to browse and install servers.
- **Claude Desktop**: Features a curated extensions directory accessible via **Settings > Extensions > Browse extensions**.
- **Other clients** (ChatGPT, Cursor, LM Studio): Currently require manual configuration with server URLs or config file edits.

## Configuration Options

The published metadata exposes a `--tools` flag that allows developers to choose between standard tools (`all`) or [code mode](https://replicate.com/changelog/2025-11-15-mcp-code-mode) (`code`) when installing the Replicate MCP server, providing flexibility in how the integration is configured.

---

### Hydra joins Supabase
- Published: 2026-02-10T07:00:00+00:00
- Products: Supabase
- Tags: release, feature, platform, integration, open-source
- URL: https://standarddb.com/supabase/supabase-acquires-hydra-team-to-build-open-data-warehouse-for-postgres--sqvqnsjm
- Original: https://supabase.com/blog/hydra-joins-supabase

## Hydra Joins Supabase

Supabase has brought on Joe Sciarrino, co-creator of Hydra, to lead the development of Supabase Warehouse and an **Open Warehouse Architecture** initiative. This move signals Supabase's commitment to making Postgres the centerpiece of modern analytics workflows.

### pg_duckdb: 600x Analytics Performance

The acquisition is built around **pg_duckdb**, an open-source MIT-licensed Postgres extension co-developed by Hydra that delivers exceptional performance gains. According to benchmarks, pg_duckdb accelerates analytics queries on Postgres by over **600x**, making it possible to run complex analytical workloads directly on Postgres without exporting data to separate systems.

### Open Warehouse Architecture

Supabase is building an open warehouse architecture that keeps Postgres at the center while unlocking modern analytics capabilities. The initiative includes:

- **Tighter Postgres + object storage integration** for seamless data pipeline workflows
- **Serverless analytics workflows** that feel native to the Supabase ecosystem
- **Open table formats support** (like Iceberg, which Supabase previously announced via Analytics Buckets)
- **Analytics-first tooling** built on open standards

This complements Supabase's earlier announcement of Analytics Buckets for storing data in Iceberg format on S3.

### Commitment to Open Source

All work remains open source. Supabase will contribute to pg_duckdb maintenance in collaboration with MotherDuck and assume maintenance of Hydra's existing repositories. Development happens in the open as the team builds out the Postgres + Analytics vision.

### Next Steps

Supabase is assembling a team to build this infrastructure and is actively hiring C++ programmers and storage engineers to work on the Open Warehouse Architecture.

---

### LangSmith is Now Available in Google Cloud Marketplace
- Published: 2026-02-10T02:47:44+00:00
- Products: 
- Tags: integration, platform, feature
- URL: https://standarddb.com/langchain/langsmith-now-available-in-google-cloud-marketplace-with-consolidated-billing--jvmhgsvg
- Original: https://blog.langchain.com/langsmith-is-now-available-in-google-cloud-marketplace

## LangSmith Lands in Google Cloud Marketplace

LangSmith, LangChain's agent engineering platform, is now available for procurement through Google Cloud Marketplace. This integration allows Google Cloud customers to purchase LangSmith directly through their existing GCP accounts, simplifying procurement workflows and enabling consolidated billing across vendors.

## Key Benefits for GCP Customers

**Billing and Procurement:**
- All LangSmith charges consolidate on existing Google Cloud invoices
- Purchases count toward Google Cloud committed spend, maximizing existing investments
- Simplified procurement with GCP already an approved vendor

**Deep Google Cloud Integration:**
LangSmith now offers native integrations across the GCP ecosystem:
- **AI Models**: First-class support for Gemini 3 Pro and Flash, plus 200+ models in Vertex AI Model Garden (Claude, Llama, Mistral)
- **Data & Storage**: Connections to AlloyDB, BigQuery, BigTable, Spanner, Firestore, and Memorystore
- **Productivity Tools**: Query Google Finance, access Google Drive documents, and analyze Google Trends data

## Platform Capabilities

LangSmith provides four core capabilities for building production AI applications:

- **Observability**: Detailed tracing and real-time monitoring with alerts and trend tracking for agent behavior auditing
- **Evaluation**: Pre-deployment testing, continuous production feedback, experiment comparison, and human annotation workflows
- **Deployment**: One-click GitHub integration with 30+ API endpoints, supporting SaaS, hybrid, and fully self-hosted deployment options
- **Agent Builder**: No-code agent creation with templates, app connections, and built-in approval workflows accessible via chat or Slack

## Getting Started

GCP customers can procure LangSmith directly from [Google Cloud Marketplace](https://console.cloud.google.com/marketplace/product/langchain-public/langsmith) or contact LangChain's sales team for enterprise requirements.

---

### Track additional Dependabot configuration changes in audit logs
- Published: 2026-02-10T00:00:00+00:00
- Products: GitHub Dependabot, GitHub
- Tags: feature, api, security, platform
- URL: https://standarddb.com/github/github-dependabot-audit-logs-now-track-configuration-changes-for-security-and--clcqghtw
- Original: https://github.blog/changelog/2026-02-10-track-additional-dependabot-configuration-changes-in-audit-logs

GitHub has expanded its audit logging capabilities for Dependabot by introducing two new event types that track configuration changes across organizations and enterprises.

## New Audit Log Events

The two new event types are:

- **Dependabot vulnerability updates toggle** (`dependabot_security_updates`): Logs when someone enables or disables Dependabot vulnerability updates on a repository
- **Self-hosted runner configuration** (`repository_dependency_updates_self_hosted`): Logs when someone enables or disables Dependabot on self-hosted runners

Each event captures critical audit information including the **actor** who made the change and the **timestamp** of when it occurred.

## Where to Find These Events

These new audit log entries are available in:
- Organization audit logs
- Enterprise audit logs

Both event types can be reviewed through GitHub's standard audit log interface for tracking configuration changes over time.

## Use Cases for Development and Security Teams

The new audit events enable organizations to:

- **Track configuration changes** for compliance and auditing purposes
- **Identify unauthorized modifications** to security settings
- **Perform forensic investigations** when needed to understand the history of Dependabot configuration changes

This enhancement provides better visibility into dependency management and security settings changes, helping teams maintain governance and compliance requirements around supply chain security.

---

### Faster MoE Training
- Published: 2026-02-10T00:00:00+00:00
- Products: 
- Tags: feature, performance, sdk, open-source
- URL: https://standarddb.com/unsloth/unsloth-ships-moe-training-kernels-with-12x-speedup-and-35-lower-vram-usage--wnflpyjs
- Original: https://unsloth.ai/docs/new/faster-moe

## MoE Training Performance Breakthrough

Unsloth released significant optimizations for training Mixture of Experts language models, achieving **~12x faster training** compared to Transformers v4, with **>35% lower VRAM usage** and support for **~6x longer context windows**. The improvements come through a combination of custom Triton kernels and a novel "Split LoRA" approach that maintains full accuracy.

## Supported Models and Hardware

The platform now supports fast MoE training for major models including:
- **Qwen3** (30B, 235B, VL, Coder variants)
- **DeepSeek R1 and V3** 
- **GPT-OSS** (20B, 120B, 500K context versions)
- **GLM** (4.6, 4.7, Flash)

Notably, gpt-oss-20b now fine-tunes on just **12.8 GB VRAM**, and the kernels work across data-center GPUs (B200, H100), consumer hardware (RTX 3090), and older generations, with support for FFT, LoRA, and QLoRA training methods.

## Technical Innovations

### Automatic Backend Selection
Unsloth automatically selects the optimal backend for your hardware:
- **grouped_mm**: PyTorch's native `torch._grouped_mm` function (optimized for H100s+, works on T4 through B200)
- **unsloth_triton**: Custom Triton kernels achieving **2.5× speedup** over grouped_mm on A100s, with auto-tuning that adds ~2 minutes overhead but yields 35% faster training runs
- **native_torch**: Fallback for compatibility, though 12x slower

### Split LoRA Approach
The key innovation is the Split LoRA method for efficient MoE training, which reduces memory consumption by ~35% and delivers 2x faster training compared to Transformers v5 + `torch._grouped_mm`.

## Integration with Transformers

This work was done in collaboration with Hugging Face, standardizing MoE training with PyTorch's `torch._grouped_mm`. Transformers v5 reorganized expert weights from a `ModuleList` of individual experts to a single `nn.Parameter`, enabling grouped matrix multiplication for dramatically faster computation. Unsloth's custom kernels push performance even further on compatible hardware.

## Getting Started

Users can upgrade via `pip install --upgrade unsloth unsloth_zoo` and access pre-built Colab notebooks for various models and configurations. Backend selection can be manually overridden via environment variables (`UNSLOTH_MOE_BACKEND`). Note: 4-bit QLoRA training for MoE models is not currently recommended due to BitsandBytes limitations; use bf16 for now.

---

### Skills support in Responses API
- Published: 2026-02-10T00:00:00+00:00
- Products: OpenAI
- Tags: 
- URL: https://standarddb.com/openai/skills-support-in-responses-api--wyyvtsdl
- Original: https://developers.openai.com/api/docs/guides/tools-skills



---

### WAF - WAF Release - Scheduled changes for 2026-02-16
- Published: 2026-02-10T00:00:00+00:00
- Products: Cloudflare
- Tags: security, feature, release
- URL: https://standarddb.com/cloudflare/cloudflare-waf-adds-detection-rules-for-zimbra-and-vite-vulnerabilities--cvmqkmzk
- Original: https://developers.cloudflare.com/changelog/scheduled-waf-release

## New WAF Detection Rules

Cloudflare will release two new detection rules for its Web Application Firewall (WAF) on **February 16, 2026**:

1. **Zimbra - Local File Inclusion (CVE-2025-68645)**: Detects exploitation attempts against a local file inclusion vulnerability in Zimbra
2. **Vite - WASM Import Path Traversal (CVE-2025-31125)**: Detects path traversal attacks targeting Vite's WASM import functionality

## Rollout Details

Both rules will launch in **log mode**, meaning they will identify suspicious traffic patterns and log the activity without blocking requests. This allows organizations to monitor for these specific threats and understand the impact before enabling active blocking rules.

## Action Items

- Review WAF rule logs starting February 16 to monitor for attempted exploitation of these vulnerabilities
- Plan to enable blocking rules once you've assessed traffic patterns in your environment
- Ensure your Zimbra and Vite installations are updated with the latest security patches

---

### WAF - WAF Release - 2026-02-10
- Published: 2026-02-10T00:00:00+00:00
- Products: Cloudflare
- Tags: bugfix, security, platform
- URL: https://standarddb.com/cloudflare/cloudflare-disables-fake-google-bot-detection-rule-in-waf--dtlnlysk
- Original: https://developers.cloudflare.com/changelog/2026-02-10-waf-release

## WAF Rule Status Change

Cloudflare has updated its Web Application Firewall (WAF) Managed Ruleset to change the default action for one security rule.

### What Changed

The **Anomaly:Header:User-Agent - Fake Google Bot** rule (ID: `...6aa0bef8`) has been transitioned from an enabled blocking state to disabled. This means the WAF will no longer block requests identified as fake Google Bot traffic by default.

### Impact

Sites using Cloudflare's Managed Ruleset will see this rule disabled in their WAF configuration. If you were relying on this rule to block spoofed bot traffic, you may need to manually re-enable it or implement alternative bot detection and mitigation strategies.

### Action Items

- Review your WAF configuration to confirm the rule status
- If you need continued protection against fake Google Bot requests, manually enable this rule or configure custom WAF rules
- Monitor your traffic patterns for any changes in bot traffic after this update

---

### GPT-5.2 Instant Update
- Published: 2026-02-10T00:00:00+00:00
- Products: OpenAI API, ChatGPT, OpenAI
- Tags: release, feature, api, model, platform
- URL: https://standarddb.com/openai/openai-updates-gpt-5-2-instant-launches-gpt-5-3-codex-with-25-faster-performance--bkhjgtfb
- Original: https://help.openai.com/en/articles/9624314-model-release-notes

## GPT-5.2 Instant Improvements

OpenAI has released an update to GPT-5.2 Instant in ChatGPT and the API that enhances response quality and contextual appropriateness. The model now produces more measured, grounded responses that better match conversation context, and delivers clearer, more relevant answers to how-to and advice-seeking questions by consistently placing critical information upfront.

## Introducing GPT-5.3-Codex

The headline addition is **GPT-5.3-Codex**, OpenAI's most capable agentic coding model to date. This model represents a significant architectural shift by combining the Codex and GPT-5 training stacks into a unified system that brings together best-in-class code generation, reasoning, and general-purpose intelligence. The model is approximately **25% faster** than its predecessor, sets new performance benchmarks, and marks a transition from pure code generation to an interactive **general-purpose coding agent** that developers can actively steer in real-time.

## Thinking Time Adjustments and Feature Updates

OpenAI has made ongoing adjustments to thinking time settings for GPT-5.2 Thinking in response to user preference data:
- **January 10, 2026**: Reduced Standard and Light thinking times based on user feedback preferring faster responses
- **February 3, 2026**: Made an additional small reduction to Standard thinking time
- **February 4, 2026**: Restored Extended thinking to its prior level after an inadvertent reduction

The thinking level toggle, introduced in September 2025, allows users to select the appropriate reasoning depth for their needs. Additionally, GPT-5.2 Instant received a personality system update on January 22 to be more conversational with better contextual tone adaptation.

## Model Retirement and Updates to Model Spec

OpenAI is retiring GPT-4o, GPT-4.1, GPT-4.1 mini, and OpenAI o4-mini from ChatGPT effective **February 13, 2026**. API users are not affected by this change. The company also released updates to its Model Spec (the living document outlining intended model behavior) with expanded **Under-18 Principles** that provide age-appropriate guidance for teen users (ages 13-17), emphasizing clear boundaries, reduced harmful content exposure, and encouraging involvement of guardians and trusted adults when risks arise. Previous updates to the Model Spec also strengthened mental health guidance and clarified support for people's real-world connections.

---

### Batch API support for GPT Image models
- Published: 2026-02-10T00:00:00+00:00
- Products: OpenAI API
- Tags: feature, api, integration
- URL: https://standarddb.com/openai/openai-adds-batch-api-support-for-gpt-image-models--tfxnmrrk
- Original: https://developers.openai.com/api/docs/guides/batch

## Batch API Support for GPT Image Models

OpenAI has expanded its Batch API to include support for GPT image generation models, providing developers with a cost-effective way to process multiple image generation requests asynchronously.

### What's New

The Batch API now allows developers to submit image generation requests in batches, which are processed during periods of lower demand. This feature:

- **Enables asynchronous processing** of multiple image generation tasks without waiting for real-time responses
- **Reduces costs** by leveraging off-peak processing capacity, similar to existing batch features for text models
- **Scales efficiently** for workloads requiring large volumes of generated images

### Use Cases

This capability is particularly valuable for:

- **Bulk image generation workflows** where real-time results aren't required
- **Cost optimization** for applications generating hundreds or thousands of images
- **Scheduled processing** of image generation tasks during off-peak hours
- **Content creation pipelines** that can tolerate delayed processing in exchange for lower costs

### Getting Started

Developers can begin using the Batch API with image models by:

1. Preparing requests in the required batch format (JSONL)
2. Submitting batches through the OpenAI API endpoints
3. Monitoring batch status and retrieving results when processing completes

For detailed implementation guidance, refer to the [Batch API documentation](https://developers.openai.com/api/docs/guides/batch) in the OpenAI API docs.

---

### v0.5.85
- Published: 2026-02-09T20:51:49+00:00
- Products: Sim Studio
- Tags: release, bugfix, feature
- URL: https://standarddb.com/sim/sim-studio-v0-5-85-improves-deploy-modal-ui-and-fixes-trigger-id-resolution--jylpxzdd
- Original: https://github.com/simstudioai/sim/releases/tag/v0.5.85

## Release Overview

Sim Studio has released v0.5.85, a minor update focused on improving the user experience and fixing a critical bug in the trigger system.

## Key Changes

**UI Improvements**
- Enhanced the deploy modal and terminal interface for better usability and clarity during deployment workflows

**Bug Fixes**
- Fixed ID resolution for tools operating in trigger mode, resolving issues where tool identifiers were not being properly resolved in certain trigger contexts

## Action Items

Users should update to v0.5.85 to benefit from the improved deployment UI and to ensure tools with trigger mode functionality work correctly.

---

### Automating Inference Optimizations with NVIDIA TensorRT LLM AutoDeploy
- Published: 2026-02-09T18:30:00+00:00
- Products: 
- Tags: feature, sdk, api, performance
- URL: https://standarddb.com/nvidia/nvidia-tensorrt-llm-adds-autodeploy-beta-automating-model-to-inference--mhlvwrzt
- Original: https://developer.nvidia.com/blog/automating-inference-optimizations-with-nvidia-tensorrt-llm-autodeploy

## AutoDeploy Shifts LLM Deployment from Manual to Compiler-Driven

NVIDIA has released AutoDeploy as a beta feature in TensorRT LLM, addressing a critical bottleneck in LLM deployment. Traditionally, deploying new model architectures requires significant manual engineering effort to add KV cache management, weight sharding, operation fusion, and hardware-specific optimizations. AutoDeploy eliminates this manual labor by automatically extracting computation graphs from off-the-shelf PyTorch models and applying optimizations through a compiler pipeline.

## Key Capabilities

AutoDeploy provides several core capabilities that accelerate deployment workflows:

- **Seamless model translation**: Automatically converts Hugging Face models into TensorRT LLM graphs without requiring engineers to rewrite model code
- **Separation of concerns**: Keeps PyTorch as the canonical model definition while delegating inference-specific optimizations to the compiler
- **Automated optimizations**: Applies sharding, quantization, KV cache insertion, attention fusion, CUDA graph optimization, and related transformations
- **Broad model support**: Currently supports over 100 text-to-text LLMs with early support for vision language models (VLMs) and state space models (SSMs)

## Technical Architecture

AutoDeploy uses PyTorch's `torch.export` API to capture models as standardized Torch graphs, then applies automated transformations for pattern matching and canonicalization of common building blocks like mixture of experts (MoE), attention layers, RoPE, and state-space components. The compiler then generates an inference-optimized TensorRT LLM graph that delegates runtime concerns to the TensorRT LLM runtime.

## Impact and Use Cases

This approach is particularly valuable for the "long tail" of models—new research architectures, internal variants, and fast-moving open source models where manual optimization is impractical. AutoDeploy enables **deployment at launch with competitive baseline performance**, while preserving a path for incremental optimization as models mature. NVIDIA demonstrated this capability by using AutoDeploy to support NVIDIA Nemotron models at their release.

---

### Introducing new token formats and secret scanning
- Published: 2026-02-09T13:00:00+00:00
- Products: Vercel
- Tags: security, api, feature
- URL: https://standarddb.com/vercel/vercel-introduces-prefixed-token-formats-and-automatic-secret-scanning-for--zmlflgts
- Original: https://vercel.com/changelog/new-token-formats-and-secret-scanning

## New Token and API Key Formats

Vercel has standardized authentication credentials with visual prefixes that make credential types immediately identifiable:

- **vcp** – Vercel personal access tokens
- **vci** – Vercel integration tokens
- **vca** – Vercel app access tokens
- **vcr** – Vercel app refresh tokens
- **vck** – Vercel API keys

These prefixes help developers quickly identify credential types in logs, configuration files, and code reviews.

## Automatic Secret Scanning and Revocation

Leveraging GitHub's secret scanning infrastructure, Vercel now automatically detects when API credentials are exposed in public GitHub repositories, gists, and npm packages. Upon detection:

- Exposed credentials are **automatically revoked** to prevent unauthorized account access
- Users receive **immediate notifications** about the breach
- Discovered tokens and API keys are visible in the Vercel dashboard for review

This provides an additional security layer for all Vercel and v0 users without requiring manual intervention.

## Recommended Actions

Developers should:

- Regularly review tokens and API keys in their Vercel account settings
- Rotate long-lived credentials periodically
- Revoke any unused authentication credentials
- Refer to Vercel's [account security documentation](https://vercel.com/docs/accounts) for best practices

---

### Bringing ChatGPT to GenAI.mil
- Published: 2026-02-09T11:00:00+00:00
- Products: ChatGPT, OpenAI
- Tags: release, platform, integration, security
- URL: https://standarddb.com/openai/openai-deploys-chatgpt-to-pentagon-s-genai-mil-platform-for-3-million-personnel--lmbbvtyf
- Original: https://openai.com/index/bringing-chatgpt-to-genaimil

## ChatGPT Arrives on DoD's GenAI.mil Platform

OpenAI announced today that ChatGPT is now available on GenAI.mil, the Department of Defense's secure enterprise AI platform serving 3 million military and civilian personnel. This deployment builds on OpenAI's existing government work, including past collaboration with DARPA on cybersecurity and a pilot program with the Department's Chief Digital and Artificial Intelligence Office (CDAO).

## Custom Deployment with Enhanced Security

The ChatGPT instance deployed on GenAI.mil is a custom version specifically configured for unclassified DoD work. Key features include:

- **Isolation and data protection**: Data processed on the platform remains isolated within government cloud infrastructure and is not used to train or improve OpenAI's commercial models
- **Built-in safeguards**: Safety controls are embedded at both the model and platform levels to ensure reliability and robustness
- **Authorized infrastructure**: The system runs on authorized government cloud infrastructure with security protections for sensitive department data

## Supported Use Cases

ChatGPT on GenAI.mil is designed to support mission-critical workflows, including:

- Summarizing and analyzing policy and guidance documents
- Drafting and reviewing procurement and contracting materials
- Generating internal reports and compliance checklists
- Supporting research, planning, mission support, and administrative tasks

OpenAI emphasizes that this deployment reflects a "focused, practical, safety-forward" approach to government AI, with the goal of helping service members access reliable AI tools while maintaining the highest standards of data protection and system reliability.

---

### Testing ads in ChatGPT
- Published: 2026-02-09T11:00:00+00:00
- Products: ChatGPT
- Tags: feature, pricing, platform
- URL: https://standarddb.com/openai/openai-begins-testing-ads-in-chatgpt-s-free-tier-keeping-conversations-private--txbfwypg
- Original: https://openai.com/index/testing-ads-in-chatgpt

## Ads Launch for Free Users

OpenAI is beginning to test advertisements in ChatGPT for logged-in adult users on the Free and Go subscription tiers in the United States. Pro, Business, Enterprise, and Education tier subscribers will not see ads. The test represents OpenAI's first implementation of advertising within its flagship product, designed to support the infrastructure costs of maintaining free access to millions of users.

## Privacy and Answer Independence

A core commitment of the test is that **ads do not influence ChatGPT's answers**. Advertisers receive only aggregated performance data (views and clicks) and have no access to user conversations, chat history, memories, or personal details. During the test, OpenAI will not show ads to accounts predicted or reported to be under 18, and ads will not appear near sensitive topics like health, mental health, or politics.

Ads are matched to conversations based on topic relevance and past interactions—for example, grocery-related ads may appear when discussing recipes—but this matching happens without sharing conversation content with advertisers. All ads are clearly labeled as sponsored and visually separated from organic answers.

## User Control and Choice

Users maintain granular control over their ad experience, including the ability to:
- Dismiss individual ads and provide feedback
- Understand why specific ads are shown to them
- Delete ad data with a single tap
- Manage personalization settings and opt out of ads entirely on the Free tier (in exchange for fewer daily free messages)

## Guiding Principles

The advertising program is built on OpenAI's previously published principles: mission alignment (funding free access through ads), answer independence, conversation privacy, user choice, and long-term value. OpenAI states its focus is learning from this initial test before any expansion, with plans to evolve the program to support additional ad formats, objectives, and business participation models.

Businesses interested in participating can sign up at openai.com/advertisers for future opportunities.

---

### R2 SQL - R2 SQL now supports approximate aggregation functions
- Published: 2026-02-09T00:00:00+00:00
- Products: Cloudflare
- Tags: feature, api, performance
- URL: https://standarddb.com/cloudflare/cloudflare-adds-approximate-aggregation-functions-to-r2-sql-for-faster-large--fscqyjxt
- Original: https://developers.cloudflare.com/changelog/2026-02-09-approximate-aggregation-functions

## New Approximate Aggregation Functions

Cloudflare has expanded R2 SQL's query capabilities with five new approximate aggregation functions designed to speed up analysis of large datasets. These functions use probabilistic algorithms to deliver fast results while accepting minimal precision trade-offs—ideal for exploratory analysis and dashboards where exact values aren't critical.

### Available Functions

- **APPROX_PERCENTILE_CONT** — Calculate percentiles on numeric data (integer and decimal columns)
- **APPROX_PERCENTILE_CONT_WITH_WEIGHT** — Weighted percentile calculations where rows contribute based on a weight column
- **APPROX_MEDIAN** — Quick median calculation (shorthand for the 0.5 percentile)
- **APPROX_DISTINCT** — Estimate unique value counts across any column type
- **APPROX_TOP_K** — Identify the k most frequent values with counts as JSON

### Flexible Filtering and Grouping

All functions support `WHERE` clause filtering for conditional analysis. Most support `GROUP BY` operations, enabling per-group analysis (e.g., median revenue by department). You can also combine approximate and standard aggregations in a single query for hybrid analysis workflows.

### Use Cases

These functions are particularly valuable for analyzing large tables where cardinality is high—think estimating unique customers across regions, calculating percentile-based metrics on revenue data, or finding top-k patterns in user behavior. The performance improvements make them suitable for interactive queries and real-time dashboards that don't require exact precision.

---

### February 9, 2026
- Published: 2026-02-09T00:00:00+00:00
- Products: 
- Tags: release, feature, api, platform, integration
- URL: https://standarddb.com/elevenlabs/elevenlabs-reorganizes-into-platform-structure-launches-custom-guardrails-and--wclhjmzt
- Original: https://elevenlabs.io/docs/changelog/2026/2/9

## Platform Reorganization

ElevenLabs is transitioning from a single-product perception to a **platform-based structure** with three clearly defined product families:

- **ElevenAgents** (Formerly Agents Platform)
- **ElevenCreative** (Formerly Creative Platform)
- **ElevenAPI** (New developer platform)

This restructuring is now reflected across documentation and SDK readmes.

## Global Server Routing Now Default

**Global routing has graduated from beta to become the default.** Previously, the default API server was located in the United States with an opt-in beta for Netherlands or Singapore routing. Now geographic proximity automatically determines server selection to optimize latency.

The deprecated `api-global-preview.elevenlabs.io` base URL should be replaced with the standard `api.elevenlabs.io`. SDK users are already using the correct default.

## Agents Platform Enhancements

The Agents Platform includes several powerful new capabilities:

- **Custom Guardrails**: Define user-configured output guardrails with custom content filtering rules beyond standard moderation. Configure with a name, prompt instruction (up to 10,000 characters), and choice of evaluation model (`gemini-2.5-flash-lite` or `gemini-2.0-flash`). Triggered guardrails end conversations.
- **WhatsApp Outbound Messaging**: New endpoint (`POST /v1/convai/whatsapp/outbound-message`) enables initiating conversations via WhatsApp using message templates.
- **Eleven v3 Conversational Model**: Added to available TTS models for agents, providing improved voice quality and expressiveness.
- **Suggested Audio Tags**: New `suggested_audio_tags` field (up to 20 tags) guides expressive speech generation for v3 models.
- **Tool Error Handling**: Configure error handling modes (`auto`, `summarized`, `passthrough`, `hide`) for webhook tools.
- **Dynamic Variable Sanitization**: New `sanitize` field removes assignment values from tool responses and transcripts while processing them for variable assignment.
- **Turn Detection Versioning**: Added `TurnModel` enum with `turn_v2` and `turn_v3` options.

## Text-to-Speech and Studio Updates

- **TTS Normalizer v3.1** improves accuracy and reduces latency for text-to-speech conversion.
- **Studio** now includes a new endpoint to retrieve muted tracks within projects (`GET /v1/studio/projects/{project_id}/muted-tracks`).

## SDK Updates

Multiple SDKs have been released with support for new features:
- **Python SDK v2.35.0** and **JavaScript SDK v2.35.0** added custom guardrails, WhatsApp outbound messaging, and related features.
- **React/Client SDKs** reduced audio chunk latency from 250ms to 100ms and improved option handling.

## Workspace and Admin Features

Workspaces now support a new **`workspace_lite_member`** seat type for limited-access members, and content templates are now shareable workspace resources.

---

### AI Search - AI Search now with more granular controls over indexing
- Published: 2026-02-09T00:00:00+00:00
- Products: Cloudflare
- Tags: feature, platform
- URL: https://standarddb.com/cloudflare/cloudflare-ai-search-adds-granular-indexing-controls-for-faster-updates--kbxplhrw
- Original: https://developers.cloudflare.com/changelog/2026-02-09-indexing-improvements

## Granular Indexing Controls

Cloudflare AI Search has introduced more fine-grained controls over the indexing process, eliminating the need for full rescans when only specific content has changed. Developers can now manage their indexed content with greater precision and efficiency.

### Reindex Individual Files

Previously, any update to indexed content required a complete rescan of the entire data source. Now, when you know exactly which file has changed or needs to be retried, you can reindex it directly without triggering a full sync.

- Navigate to **Overview** > **Indexed Items**
- Select the sync icon next to any file to reindex it immediately
- Changes take effect without rescanning the entire data source

### Selective Sitemap Crawling

For websites with multiple sitemaps, you can now specify which sitemap the crawler should use instead of processing all sitemaps listed in `robots.txt`. This is particularly useful for sites with separate documentation, blog, and product sitemaps.

- Go to **Settings** > **Parsing options** > **Specific sitemaps**
- Enter the URL of the sitemap you want to crawl (e.g., `https://example.com/docs-sitemap.xml`)
- Trigger a sync to apply the changes
- The crawler will respect your selection up to the [maximum files per index limit](https://developers.cloudflare.com/ai-search/platform/limits-pricing/#limits)

These improvements reduce indexing overhead and give you better control over what content is included in your AI Search index, resulting in faster updates and more relevant search results.

---

### Transformers.js v4 Preview: Now Available on NPM!
- Published: 2026-02-09T00:00:00+00:00
- Products: 
- Tags: release, feature, sdk, performance, open-source
- URL: https://standarddb.com/huggingface/transformers-js-v4-preview-debuts-on-npm-with-new-webgpu-runtime-and-10x-build--qnmpkyhw
- Original: https://huggingface.co/blog/transformersjs-v4

## Major Runtime Rewrite

Transformers.js v4 features a completely rebuilt WebGPU runtime written in C++ and developed in collaboration with the ONNX Runtime team. This new foundation enables the same transformer inference code to run across diverse JavaScript environments—browsers, Node.js, Bun, and Deno—all with GPU acceleration support.

## Performance Optimizations

The v4 release delivers significant performance gains through strategic use of ONNX Runtime's Contrib Operators. By leveraging specialized operators like `com.microsoft.GroupQueryAttention`, `com.microsoft.MatMulNBits`, and `com.microsoft.MultiHeadAttention`, the team achieved **~4x speedups for BERT-based embedding models**. The build system migration from Webpack to esbuild reduced build times from 2 seconds to 200 milliseconds (**10x improvement**) and decreased average bundle sizes by 10%.

## Expanded Model Support

v4 introduces support for new model architectures including GPT-OSS, Chatterbox, GraniteMoeHybrid, HunYuanDenseV1, Olmo3, and Youtu-LLM. The release adds first-class support for advanced patterns like Mamba (state-space models), Multi-head Latent Attention (MLA), and Mixture of Experts (MoE), bringing the total supported architectures to ~200.

## Code Quality and Maintainability

The development cycle included major refactoring efforts: conversion to a pnpm monorepo workspace for modular sub-packages, splitting the 8,000+ line models.js file into focused modules, moving examples to a dedicated repository, and adopting Prettier for consistent formatting. These changes reduce technical debt and lower the barrier for contributors adding new models.

## Getting Started

Install the preview version with:
```sh
npm i @huggingface/transformers@next
```

The team will continue publishing regular v4 updates under the `next` tag on NPM before the full release.

---

### Agents - Interactive browser terminals in Sandboxes
- Published: 2026-02-09T00:00:00+00:00
- Products: Cloudflare Workers
- Tags: feature, sdk, api, release
- URL: https://standarddb.com/cloudflare/cloudflare-sandbox-sdk-adds-pty-terminal-support-for-browser-based-shell-access--vpkkndld
- Original: https://developers.cloudflare.com/changelog/2026-02-09-pty-terminal-support

## Interactive Browser Terminals Now Available

The Sandbox SDK has been enhanced with **PTY (pseudo-terminal) passthrough** capabilities, enabling developers to build browser-based terminal interfaces that directly connect to sandbox container shells via WebSocket.

### Key Features

**New `sandbox.terminal()` method**: Proxies WebSocket upgrades to the container's PTY endpoint with built-in output buffering for seamless reconnection playback.

**Multiple isolated terminals**: Each session can maintain its own terminal with independent working directories and environments, allowing users to run separate shells side-by-side within the same container.

**xterm.js integration**: A new `@cloudflare/sandbox/xterm` addon provides:
- Automatic reconnection with exponential backoff and jitter
- Buffered output replay on reconnect
- Automatic resize forwarding to match terminal dimensions

### Getting Started

Update to the latest version with:
```
npm i @cloudflare/sandbox@latest
```

The simple API allows you to expose a terminal from your Worker:
```javascript
return sandbox.terminal(request, { cols: 80, rows: 24 });
```

For xterm.js users, the new addon handles connection management and state updates automatically, reducing boilerplate code for interactive terminal UIs.

---

### AI Crawl Control - Analytics enhancements
- Published: 2026-02-09T00:00:00+00:00
- Products: Cloudflare
- Tags: feature, platform
- URL: https://standarddb.com/cloudflare/cloudflare-ai-crawl-control-adds-path-grouping-referral-analytics-and-data--nctpjylj
- Original: https://developers.cloudflare.com/changelog/2026-02-09-analytics-enhancements

## AI Crawl Control Analytics Enhancements

Cloudflare has released several improvements to the AI Crawl Control analytics dashboard to provide better visibility into AI crawler activity and its impact on your infrastructure.

### Path Pattern Grouping

The **Metrics** tab now includes a **Patterns** view that automatically groups requests by URI patterns (e.g., `/blog/*`, `/api/v1/*`, `/docs/*`). This aggregated view helps you quickly identify which site areas are being targeted most heavily by AI crawlers without needing to parse individual URLs.

### Enhanced Referral Analytics

- **Destination patterns** now show which site areas receive AI-driven referral traffic
- A new **Referrals over time** chart in the **Metrics** tab displays trends broken down by operator or source, allowing you to track traffic patterns over time

### Data Transfer Visibility

Two new metrics help you understand bandwidth impact:
- Toggle **Bytes** in the **Allowed requests over time** chart to visualize bandwidth consumption alongside request volume
- A new **Bytes Transferred** column in the **Crawlers** tab shows per-crawler bandwidth usage

### Image Exports

You can now export charts and tables as images, simplifying the creation of reports and presentations for stakeholders.

For detailed guidance on using these features, refer to the [analyzing AI traffic documentation](https://developers.cloudflare.com/ai-crawl-control/features/analyze-ai-traffic/).

---

### GPT-5.3-Codex is now generally available for GitHub Copilot
- Published: 2026-02-09T00:00:00+00:00
- Products: GitHub Copilot
- Tags: release, model, feature
- URL: https://standarddb.com/github/github-copilot-launches-gpt-5-3-codex-with-25-faster-performance-on-coding-tasks--tvqstzjn
- Original: https://github.blog/changelog/2026-02-09-gpt-5-3-codex-is-now-generally-available-for-github-copilot

## GPT-5.3-Codex Now Available in GitHub Copilot

GitHub has rolled out GPT-5.3-Codex, OpenAI's latest agentic coding model, to all GitHub Copilot users. Early testing shows the model reaches new high scores on coding, agentic, and real-world capability benchmarks, with **up to 25% faster performance than GPT-5.2-Codex on agentic coding tasks**.

## Where to Access

The model is available to Copilot Pro, Pro+, Business, and Enterprise users through:
- Visual Studio Code (chat, ask, edit, and agent modes)
- github.com
- GitHub Mobile (iOS and Android)
- GitHub CLI
- GitHub Copilot Coding Agent

Users can select GPT-5.3-Codex from the model picker in their preferred environment. Rollout is gradual, so some users may see the option available shortly.

## Administrator Configuration

Copilot Enterprise and Copilot Business plan administrators need to enable the GPT-5.3-Codex policy in Copilot settings to grant access to their teams.

## What's Improved

The new model features enhanced reasoning and execution capabilities specifically designed for complex, tool-driven, long-running workflows. This makes it particularly well-suited for agentic tasks that require multi-step problem solving and iterative refinement.

---

### GitHub Apps can now utilize public preview Enterprise Teams APIs via fine-grained permissions
- Published: 2026-02-09T00:00:00+00:00
- Products: GitHub
- Tags: api, feature, platform
- URL: https://standarddb.com/github/github-apps-gain-fine-grained-permissions-for-enterprise-teams-api-management--gkmlzvgk
- Original: https://github.blog/changelog/2026-02-09-github-apps-can-now-utilize-public-preview-enterprise-teams-apis-via-fine-grained-permissions

## What Changed

GitHub Enterprise administrators now have the ability to use GitHub Apps with **enterprise teams** fine-grained permissions to access the Enterprise Teams API endpoints. Previously, these endpoints required personal access tokens (classic), which presented security and management challenges for enterprises.

## New Capabilities

The new `enterprise teams` fine-grained permission provides two access levels:

- **Read access**: Retrieve enterprise team information programmatically
- **Write access**: Create, update, and delete operations for enterprise-level team management

This enables enterprise administrators to securely manage teams at scale through programmatic app integrations, offering better security posture and easier access control compared to managing personal access tokens.

## Developer Action Items

If you're building GitHub Apps for enterprise customers:

1. Review the [GitHub Apps authentication documentation](https://docs.github.com/enterprise-cloud@latest/apps/creating-github-apps/authenticating-with-a-github-app/about-authentication-with-a-github-app) for implementation details
2. Consult the [Enterprise Teams API endpoints reference](https://docs.github.com/enterprise-cloud@latest/rest/enterprise-teams?apiVersion=2022-11-28) to understand available operations
3. Update your app configurations to use the new fine-grained `enterprise teams` permission instead of requiring personal access tokens from users
4. Consider joining the [discussion thread](https://github.com/orgs/community/discussions/177040) to provide feedback or ask questions

---

### Agents, Workers - Agents SDK v0.4.0: Readonly connections, MCP security improvements, x402 v2 migration, and custom MCP OAuth providers
- Published: 2026-02-09T00:00:00+00:00
- Products: Cloudflare Workers
- Tags: release, feature, api, security, breaking-change, sdk
- URL: https://standarddb.com/cloudflare/cloudflare-agents-sdk-v0-4-0-adds-readonly-connections-mcp-oauth-customization--yvvbxqyy
- Original: https://developers.cloudflare.com/changelog/2026-02-09-agents-sdk-v040

## Key Features

**Readonly Connections** allow agents to restrict WebSocket clients to read-only access, blocking both client-side `setState()` calls and mutating `@callable()` methods. This is essential for dashboards and spectator views where observers should not modify agent state. New hooks—`shouldConnectionBeReadonly`, `setConnectionReadonly`, and `isConnectionReadonly`—enable fine-grained control, with the readonly flag persisting across hibernation.

**Custom MCP OAuth Providers** give developers the ability to override the default OAuth provider used when connecting to MCP servers. The new `createMcpOAuthProvider` method on the Agent class enables custom authentication strategies like pre-registered client credentials or mTLS, beyond the built-in dynamic client registration.

## Security and Stability Improvements

The MCP SDK was upgraded to v1.26.0 to prevent cross-client response leakage. Stateless MCP servers must now create a new `McpServer` instance per request instead of sharing a single instance. An important security fix adds a `callbackPath` option to `addMcpServer` to prevent instance name leakage in MCP OAuth callback URLs—when `sendIdentityOnConnect` is false, this parameter is now required.

## x402 v2 Migration

The x402 payment integration migrated from the legacy `x402` package to `@x402/core` and `@x402/evm` v2. **Breaking changes** include updated peer dependencies, new `PaymentRequirements` fields (e.g., `amount` instead of `maxAmountRequired`), and changed `X402ClientConfig.account` type. Network identifiers now accept both legacy names and CAIP-2 format, with automatic conversion support. Server-side lazy initialization defers facilitator connection until the first paid tool invocation, and payment tokens support both v2 and v1 HTTP headers.

## API Changes and Deprecations

`onStateUpdate` is deprecated in favor of `onStateChanged` (a drop-in rename with identical behavior). The `onStateUpdate` hook still works but emits a one-time console warning. A new client-side `onStateUpdateError` callback handles rejected state updates, and `validateStateChange` rejections now propagate a `CF_AGENT_STATE_ERROR` message back to the client. CORS handling was simplified by delegating to partyserver's native support.

To upgrade:
```bash
npm i agents@latest
```

---

### Bun v1.3.9
- Published: 2026-02-08T09:01:15+00:00
- Products: 
- Tags: release, feature, performance, bugfix
- URL: https://standarddb.com/bun/bun-v1-3-9-adds-parallel-sequential-script-running-http-2-upgrades-and-28--yjfvzttb
- Original: https://bun.com/blog/bun-v1.3.9

## Key Features

**Parallel and Sequential Script Running**

The new `bun run --parallel` and `bun run --sequential` commands let you execute multiple package.json scripts concurrently or sequentially with colored, prefixed output. These commands integrate fully with `--filter` and `--workspaces` for coordinating builds across monorepo packages:

```
bun run --parallel build test
bun run --sequential --workspaces build
bun run --parallel --filter '*' build lint test
```

Unlike `--filter`, which respects dependency order, these new flags start all scripts immediately without waiting for dependencies—making them ideal for long-lived watch processes. Use `--no-exit-on-error` to continue even if one script fails.

**HTTP/2 Connection Upgrade Support**

Bun now correctly handles the `net.Server → Http2SecureServer` connection upgrade pattern. Libraries like [http2-wrapper](https://github.com/szmarczak/http2-wrapper) and [crawlee](https://github.com/apify/crawlee) that emit raw TCP connections to HTTP/2 servers will now work properly out of the box.

**Testing Improvements**

`mock()` and `spyOn()` now implement `Symbol.dispose`, enabling the `using` keyword for automatic cleanup:

```
using spy = spyOn(obj, "method").mockReturnValue("mocked");
// automatically restored when leaving scope
```

## Fixes and Optimizations

- **NO_PROXY respected for explicit proxy options**: Previously ignored when proxies were explicitly passed; now always checked
- **ARM64 compatibility**: Fixed illegal instruction crashes on ARMv8.0 processors (Cortex-A53, Raspberry Pi 4, AWS a1) by correctly targeting ARMv8.0 instead of requiring ARMv8.1+
- **Markdown rendering**: 3-15% faster HTML rendering via SIMD-accelerated character scanning
- **React markdown**: 28% faster small document rendering through cached HTML tag strings
- **ESM bytecode support**: `--bytecode` now works with `--format=esm`
- **CPU profiler**: Added `--cpu-prof-interval` flag matching Node.js for configurable sampling intervals

---

### v0.15.6
- Published: 2026-02-08T05:00:05+00:00
- Products: 
- Tags: release, bugfix
- URL: https://standarddb.com/ollama/ollama-v0-15-6-fixes-context-limits-and-adds-automatic-model-downloads--ndndyvbz
- Original: https://github.com/ollama/ollama/releases/tag/v0.15.6

## Key Fixes

Ollama v0.15.6 addresses several important bugs that improve the reliability and usability of the platform:

**Context Management Improvements**
- Fixed context limits when running `ollama launch droid`
- Resolved an issue where `ollama launch claude` would cause context compaction when providing images

**Automatic Model Downloads**
- `ollama launch` now automatically downloads missing models instead of returning an error, reducing friction when users try to run models they haven't yet installed

## What Developers Should Know

These are targeted bugfixes focused on improving the launcher experience and ensuring models handle context windows correctly. Users running the droid model or using Claude with image inputs should update to resolve these issues. The automatic model download feature makes the launch command more forgiving and user-friendly.

## Update Guidance

Update to v0.15.6 to take advantage of these fixes, particularly if you've encountered context-related issues or manual model download requirements when using `ollama launch`.

---

### v0.5.84
- Published: 2026-02-08T03:17:04+00:00
- Products: Sim Studio
- Tags: release, bugfix, security
- URL: https://standarddb.com/sim/sim-v0-5-84-released-with-request-sanitization-security-fix--lwkjrvcg
- Original: https://github.com/simstudioai/sim/releases/tag/v0.5.84

## Bug Fixes

**Request Sanitization Enhancement**

Sim v0.5.84 introduces improved request sanitization in the models component. This security-focused update addresses potential input validation vulnerabilities that could impact model request handling.

## What Changed

- Added request sanitization to the models component ([#3165](https://github.com/simstudioai/sim/pull/3165))

## Action Items

Users should upgrade to v0.5.84 to ensure their instances benefit from the latest security hardening. The update is a straightforward patch release with no breaking changes or new features.

## Release Details

- **Version**: v0.5.84
- **Release Date**: February 7, 2026
- **Contributors**: @waleedlatif1

---

### v0.5.83
- Published: 2026-02-07T20:36:56+00:00
- Products: Sim Studio
- Tags: release, feature, bugfix, platform
- URL: https://standarddb.com/sim/sim-v0-5-83-adds-skills-to-agent-blocks-airweave-integration-and-improved-azure--fftfkxls
- Original: https://github.com/simstudioai/sim/releases/tag/v0.5.83

## New Features

This release introduces three significant feature additions to the Sim Studio platform:

- **Skills for Agent Blocks**: Agent blocks now support skills, enabling more sophisticated agent behavior and capabilities within workflows.
- **Airweave Block Integration**: A new Airweave block has been added, expanding integration options for workflow orchestration.
- **Enhanced Model Configuration**: Updated model configs and improved Anthropic provider error handling to propagate errors back to users.

## Improvements and Fixes

**UI and User Experience**:
- Reordered models dropdown for better discoverability
- Enhanced preview rendering for nested values like input formats in workflow execution
- Improved skills UI, validation, and permissions management

**Backend and Infrastructure**:
- Fixed Azure-Anthropic support across router, evaluator, copilot, and tokenization components
- Fixed isolated-vm worker pool to prevent single-worker bottlenecks and improved execution user ID resolution
- Resolved edge case handling for 1-hour TTL in rooms
- Swapped out hybrid auth at relevant callsites for more consistent authentication

## What Developers Should Know

These improvements enhance both the capabilities and stability of Sim Studio workflows. The addition of skills to agent blocks and Airweave integration provide more building blocks for complex automation. Bug fixes address critical infrastructure issues that could impact workflow execution performance and reliability.

---

### Opus 4.6 (fast mode) is now available in Windsurf
- Published: 2026-02-07T12:00:00+00:00
- Products: 
- Tags: 
- URL: https://standarddb.com/windsurf/opus-4-6-fast-mode-is-now-available-in-windsurf--xcdfvnrt
- Original: https://windsurf.com/blog/opus-4.6-fast



---

### Fast mode for Claude Opus 4.6 is now in preview for GitHub Copilot
- Published: 2026-02-07T00:00:00+00:00
- Products: GitHub Copilot
- Tags: release, feature, model, performance
- URL: https://standarddb.com/github/claude-opus-4-6-fast-mode-launches-in-github-copilot-preview-with-2-5x-faster--pmfhshgn
- Original: https://github.blog/changelog/2026-02-07-claude-opus-4-6-fast-is-now-in-public-preview-for-github-copilot

## Fast Mode Performance Boost

GitHub is rolling out a research preview of fast mode for Claude Opus 4.6 in GitHub Copilot. This optimized variant delivers significantly accelerated inference, with output token speeds up to **2.5x faster** than standard Opus 4.6, while maintaining identical intelligence and capabilities. This is an early, experimental release focused on speed optimization for development workflows.

## Availability and Access

Fast mode for Claude Opus 4.6 is available to **Copilot Pro+ and Enterprise** subscribers. Users can select the model via the model picker in:

- Visual Studio Code (chat, ask, edit, and agent modes)
- Copilot CLI

The rollout is gradual—if the model doesn't appear immediately, check back soon as deployment continues.

## Important Pricing Note

**Editor's note:** The preview promotional period ends February 16, 2025 (Pacific Time). After the promotion expires, a **30x premium request multiplier** will apply to this model, significantly increasing usage costs.

## For Enterprise Administrators

Copilot Enterprise plan administrators must manually enable the Fast mode for Claude Opus 4.6 policy in Copilot settings to grant access to their teams.

## Next Steps

Developers should test fast mode for their specific use cases and provide feedback in the [GitHub Community discussions](https://github.com/orgs/community/discussions/categories/copilot-conversations) to help shape further improvements.

---

### The New v0 Is Ready for Production Apps and Agents
- Published: 2026-02-06T20:45:28+00:00
- Products: 
- Tags: release, feature, platform, integration
- URL: https://standarddb.com/neon/vercel-s-v0-moves-from-prototyping-to-production-with-real-codebases-and-git--syxmfghn
- Original: https://neon.com/blog/the-new-v0-is-ready-for-production-apps-and-agents

## Production-Ready Architecture

Vercel has fundamentally reimagined v0, shifting from an isolated code generation tool to an end-to-end development platform. The new v0 works directly against existing GitHub repositories with real environment variables, configuration, and runtime constraints—eliminating the gap between generated code and production-ready systems.

## Key Features

**Real Codebase Integration**: Import and work directly within existing repositories instead of generating code in isolation. Changes integrate seamlessly with your actual project structure.

**Git-First Workflows**: v0 now handles Git workflows natively, making pull requests, previews, and deployments accessible to non-technical team members. Each chat creates a branch, each branch gets a preview environment, and merging follows standard production deployment paths.

**Enterprise Security**: Built on Vercel's production infrastructure with deployment protections, access controls, and enterprise-grade security defaults baked in from the start.

**Database Integration**: v0 integrates with database partners like Neon, allowing teams to provision isolated database branches alongside application preview environments. This enables real data testing without shared staging databases or mocks.

## Looking Ahead to Agents

This architectural redesign isn't just about static apps—it establishes foundations for agentic workflows where AI systems can operate safely across real codebases, real data, and real infrastructure with proper security boundaries and audit trails.

## Getting Started

Neon-powered templates are available at v0.app/@neonpostgres, providing starter projects that combine v0's AI-assisted development with production database infrastructure out of the box.

---

### Sanity is now available on the Vercel Marketplace
- Published: 2026-02-06T13:00:00+00:00
- Products: Next.js, Vercel
- Tags: integration, platform, feature
- URL: https://standarddb.com/vercel/sanity-cms-now-available-on-vercel-marketplace-with-native-integration--spqcdmnr
- Original: https://vercel.com/changelog/sanity-vercel-marketplace

## Native CMS Integration in Vercel Marketplace

Sanity is now available on the Vercel Marketplace as a first-class CMS integration. This addition brings streamlined content management capabilities directly into the Vercel platform, reducing setup friction and keeping your entire development workflow in one place.

## Key Benefits

- **Unified Setup**: Install, configure, and manage Sanity directly from the Vercel dashboard
- **Automatic Provisioning**: The integration automatically provisions your Sanity project ID as an environment variable
- **No Manual Configuration**: Eliminates the need for manual API token setup and environment variable configuration
- **Integrated Workflow**: Keep CMS provisioning and account management within your existing Vercel workflow instead of context-switching to a separate dashboard

## Getting Started

The integration works seamlessly with Next.js projects. You can:

1. **Define Content Schemas** in code using Sanity's schema system, specifying document types and their fields
2. **Configure the Sanity Client** with automatically provisioned environment variables
3. **Query Content** using GROQ (Graph-Relational Object Queries), Sanity's query language for precise field selection

The Marketplace also provides a ready-to-use template: the [Next.js + Sanity Personal Website](https://vercel.com/templates/template/sanity-next-js-personal-website), allowing developers to deploy a working example immediately.

To get started, [install Sanity from the Vercel Marketplace](https://vercel.com/marketplace/sanity) or deploy the provided template from your Vercel dashboard.

---

### Simplified file retrieval from Vercel Sandbox environments
- Published: 2026-02-06T13:00:00+00:00
- Products: Vercel
- Tags: feature, sdk, api
- URL: https://standarddb.com/vercel/vercel-sandbox-sdk-adds-file-retrieval-methods-for-simplified-data-extraction--pxvddjhm
- Original: https://vercel.com/changelog/simplified-file-retrieval-from-vercel-sandbox-environments

## New File Retrieval Methods

The [Vercel Sandbox SDK](https://www.npmjs.com/package/@vercel/sandbox) now includes two new methods that simplify retrieving files generated within sandbox environments:

- **`downloadFile()`** - Seamlessly streams and downloads files from the sandbox to your local machine
- **`readFileToBuffer()`** - Reads file contents directly into a buffer without manual stream operations

## What's Changed

Previously, retrieving files from Vercel Sandbox environments required developers to manually handle stream operations and implement custom piping logic. Code running in sandboxes often generates output files—such as CSV reports, processed images, or PDF invoices—that need to be transferred across the network boundary from isolated VMs.

The new methods abstract away these low-level stream operations, providing a clean API for common file retrieval patterns.

## Usage Examples

**Download a file to your local machine:**
```typescript
const dstPath = await sandbox.downloadFile(
  { path: 'generated-file.csv', cwd: '/vercel/sandbox' },
  { path: 'generated-file.csv', cwd: '/tmp' }
);
```

**Read file contents into memory:**
```typescript
const buffer = await sandbox.readFileToBuffer({ path: 'chart.png' });
```

## Next Steps

Developers using Vercel Sandbox can immediately start leveraging these methods. Refer to the [Sandbox SDK documentation](https://www.npmjs.com/package/@vercel/sandbox) and [SDK reference guide](https://vercel.com/docs/vercel-sandbox/sdk-reference#sandbox.downloadfile) for detailed usage patterns and additional capabilities.

---

### v0.5.82
- Published: 2026-02-06T08:54:13+00:00
- Products: Sim Studio
- Tags: release, feature, bugfix, integration
- URL: https://standarddb.com/sim/sim-v0-5-82-adds-slack-file-attachments-fixes-resolver-and-integrations--srjwgqxl
- Original: https://github.com/simstudioai/sim/releases/tag/v0.5.82

## What's New

**Slack File Attachment Support**
The release adds file attachment support to Slack webhook triggers, enabling workflows to send files directly through Slack integration without additional workarounds.

## Bug Fixes

This patch release addresses several critical issues:

- **Resolver metrics**: Fixed response format and evaluator metrics in deactivated branch execution (#3152)
- **Linear integration**: Corrected tool outputs, queries, and pagination alignment with Linear's API (#3150)
- **Executor loops**: Resolved issue where loop sentinel-end was wrongly queued (#3148)

## Update Details

This is a minor incremental release that improves integration reliability and workflow execution accuracy. Users relying on Slack integrations will benefit from the new file attachment capability, while existing workflows using Linear or branch-based logic will see improved stability.

---

### X / Twitter OAuth 2.0 is now available for Supabase Auth
- Published: 2026-02-06T07:00:00+00:00
- Products: Supabase
- Tags: feature, api, integration, security
- URL: https://standarddb.com/supabase/supabase-auth-adds-x-twitter-oauth-2-0-provider-with-pkce-support--xfxtyfhw
- Original: https://supabase.com/blog/x-twitter-oauth-2-provider

## X / Twitter OAuth 2.0 Now Available

Developers can now integrate "Sign in with X" functionality into their applications using Supabase Auth's new **X / Twitter OAuth 2.0 provider**. This replaces the legacy OAuth 1.0a implementation with a modern authentication approach.

### Key Improvements

The OAuth 2.0 implementation brings several advantages:
- **Modern security practices** including PKCE (Proof Key for Code Exchange) support
- **Better authentication experience** aligned with current industry standards
- **Simplified integration** through the Supabase dashboard

### Setup Process

Getting X / Twitter authentication running takes just a few steps:

1. Create an OAuth 2.0 app in the [X Developer Portal](https://developer.x.com/en/portal/dashboard)
2. Enable "Request email from users" in your app's authentication settings
3. Configure your callback URL from the Supabase dashboard
4. Retrieve your **Client ID** and **Client Secret** from the "Keys and tokens" section
5. Enter credentials in **Authentication > Providers > X / Twitter (OAuth 2.0)** in Supabase

Once configured, use the Supabase client library to authenticate:
```javascript
const { data, error } = await supabase.auth.signInWithOAuth({
  provider: 'x',
})
```

### Migration Path

If you're currently using the legacy **Twitter (OAuth 1.0a)** provider, migration to OAuth 2.0 is recommended. The legacy provider will remain available as long as X / Twitter continues supporting OAuth 1.0a, giving teams time to transition. See the [full documentation](https://supabase.com/docs/guides/auth/social-login/auth-twitter) for complete setup guidance.

---

### v0.15.5
- Published: 2026-02-06T05:43:09+00:00
- Products: 
- Tags: release, feature, api, model, bugfix
- URL: https://standarddb.com/ollama/ollama-v0-15-5-adds-two-new-models-and-improves-agentic-coding-support--tsrdrbpt
- Original: https://github.com/ollama/ollama/releases/tag/v0.15.5

## New Models

Ollama v0.15.5 introduces two new models to the library:

- **Qwen3-Coder-Next**: A coding-focused language model from Alibaba's Qwen team, optimized for agentic coding workflows and local development environments.
- **GLM-OCR**: A multimodal OCR model for complex document understanding, built on the GLM-V encoder–decoder architecture for advanced text and layout recognition.

## Improvements to Agentic Features

The `ollama launch` command receives significant enhancements for agent-based workflows:

- **Sub-agent support**: `ollama launch` can now spawn and manage sub-agents for planning, deep research, and similar multi-step tasks.
- **Flexible arguments**: Arguments can now be passed through `ollama launch`, e.g., `ollama launch claude -- --resume`.
- **Model-specific context tuning**: Context limits are automatically set for specific models (e.g., `ollama launch opencode`).

## System Improvements

- **Automatic context length tuning**: Ollama now defaults context lengths based on available VRAM:
  - Less than 24 GiB: 4,096 tokens
  - 24–48 GiB: 32,768 tokens
  - 48 GiB or more: 262,144 tokens
- **Simplified authentication**: `ollama signin` now opens a browser window directly to the connection page.
- **MLX engine expansion**: Added support for GLM-4.7-Flash on the experimental MLX engine.

## Bug Fixes

- Fixed an off-by-one error when using the `num_predict` API parameter.
- Resolved issue where tokens from previous sequences would be incorrectly returned when hitting `num_predict` limits.



---

### Workers - Visualize data, share links, and create exports with the new Workers Observability dashboard
- Published: 2026-02-06T00:00:00+00:00
- Products: Cloudflare Workers
- Tags: feature, platform, release
- URL: https://standarddb.com/cloudflare/cloudflare-workers-observability-dashboard-adds-data-visualization-exports-and--snvmsbbd
- Original: https://developers.cloudflare.com/changelog/2026-02-06-observability-ui-refresh

## Overview

The Workers Observability dashboard now includes powerful new features to streamline debugging workflows and team collaboration. These updates are available immediately in both individual Worker Observability tabs and the account-level Observability dashboard for a consistent experience across Cloudflare's platform.

## Key Features

**Data Visualization & Analysis**
- Build custom charts directly from Worker data without leaving the dashboard
- Visualize metrics like P99 CPU time, grouped by outcome or other dimensions
- Export logs and traces as JSON or CSV for offline analysis

**Sharing & Collaboration**
- Generate direct, shareable URLs to specific events, invocations, and traces
- Standalone pages display full context when shared links are accessed
- Simplify team collaboration by reducing context-switching

**Improved User Experience**
- Customizable table columns—add, remove, and reorder fields in the events table
- Expandable event details for inline inspection without leaving the table view
- Keyboard shortcuts for faster navigation and workflow efficiency

## Getting Started

Access the enhanced Observability dashboard by navigating to **Workers & Pages** > select your Worker > **Observability** in the Cloudflare dashboard. The features are live for all users immediately.

---

### February 6, 2026
- Published: 2026-02-06T00:00:00+00:00
- Products: 
- Tags: feature, api, model, integration
- URL: https://standarddb.com/deepgram/deepgram-adds-openai-gpt-5-2-and-google-gemini-3-support-to-voice-agent-api--xybwsnzv
- Original: https://developers.deepgram.com/changelog/2026/2/6

Deepgram has expanded the LLM capabilities of its Voice Agent API by adding support for the latest models from OpenAI and Google.

**New Models Available:**
- **OpenAI GPT-5.2 Instant** (gpt-5.2-instant) — optimized for faster responses
- **OpenAI GPT-5.2 Thinking** (gpt-5.2) — designed for complex reasoning tasks
- **Google Gemini 3 Flash** (gemini-3-flash-preview) — Google's latest preview model

**How to Use:**
Developers can select any of these new models by configuring their Voice Agent settings with the appropriate provider and model name. The configuration follows Deepgram's existing settings structure, making it straightforward to integrate into existing Voice Agent implementations.

**Next Steps:**
For detailed information about all supported LLM models and configuration options, refer to the Voice Agent LLM Models documentation.

---

### What's New in Suno Studio 1.2
- Published: 2026-02-06T00:00:00+00:00
- Products: 
- Tags: release, feature
- URL: https://standarddb.com/suno/suno-studio-1-2-adds-audio-warping-time-signature-support-and-effect-removal--vrkggvjc
- Original: https://www.suno.com/blog/studio1_2

## New Audio Editing Capabilities

Suno Studio 1.2 brings several powerful features designed to give music creators more granular control over composition and production without leaving the platform.

**Remove FX** allows users to strip effects from audio clips, generating clean, effect-free versions suitable for export to Digital Audio Workstations (DAWs). This is particularly useful for vocals or instruments with heavy reverb or other processing that users want to customize independently.

**Warp Markers with Quantize** enables timing and groove adjustments through customizable markers. Users can either manually place markers throughout an audio file or automatically detect transients. The quantize feature snaps loose recordings to the grid, ensuring everything locks to the session tempo — essential for polished production.

**Alternates** streamlines the take management workflow by allowing users to create and audition multiple variations of sounds from a single track. Users can quickly preview different takes, select their favorite, and maintain organized sessions without cluttered track layouts.

## Expanded Creative Options

The addition of **Time Signature Support** eliminates the previous 4/4 limitation. Creators can now work in various time signatures including 6/8, 7/8, 11/4, and other odd meters — opening doors for diverse musical styles from traditional world music to progressive and math rock compositions.

These features position Suno Studio as a more comprehensive production tool, bridging the gap between AI-assisted generation and traditional DAW workflows for Premier subscribers.

---

### CodeQL 2.24.1 improves Maven private registry support and improves query accuracy
- Published: 2026-02-06T00:00:00+00:00
- Products: GitHub Actions, GitHub
- Tags: release, feature, security, performance, bugfix
- URL: https://standarddb.com/github/codeql-2-24-1-improves-maven-registry-support-and-query-accuracy-across-seven--twjcczlr
- Original: https://github.blog/changelog/2026-02-06-codeql-2-24-1-improves-maven-private-registry-support-and-improves-query-accuracy

## Language and Framework Support Updates

CodeQL 2.24.1 expands language coverage with support for Kotlin up to version 2.3.0, C23 and C++26 preprocessor directives, and C# 14 null-conditional assignments. Java developers benefit from added support for Struts 7.x package names. Note that support for older Kotlin versions (1.6.x and 1.7.x) has been discontinued.

## Maven Registry Enhancements

A significant improvement for enterprise environments: Maven-compatible private package registries configured for an organization now automatically extend to Maven plugin repositories. This enables teams to obtain Maven plugins from private registries when using CodeQL's Default Setup, streamlining dependency management for organizations with internal registries.

## Query Accuracy Improvements

Multiple queries receive accuracy enhancements to reduce false positives:

- **C/C++**: Fixed a `GuardCondition` library bug that prevented proper recognition of binary logical operators in guard conditions. Buffer size measurement improvements reduce false positives in overflow-related queries (`cpp/static-buffer-overflow`, `cpp/overflow-buffer`, `cpp/badly-bounded-write`, and related queries).
- **Java**: Improved accuracy in the `java/unreleased-lock` query.
- **Python**: New experimental `py/prompt-injection` query detects potential prompt injection vulnerabilities in LLM code.

## Additional Improvements

Python models-as-data now supports referring to list elements via the `ListElement` path, and new taint flow models cover the `agents` and `openai` modules. A crash affecting GitHub Actions analysis of long `${{ ... }}` expressions (around 300+ characters) has been fixed.

## Deployment

CodeQL 2.24.1 is automatically deployed to GitHub.com users. GitHub Enterprise Server users on older versions can manually upgrade their CodeQL installation.

---

### v0.5.81
- Published: 2026-02-05T19:48:08+00:00
- Products: Sim Studio
- Tags: release, feature, integration, api
- URL: https://standarddb.com/sim/sim-releases-v0-5-81-with-azure-anthropic-support-and-confluence-integrations--xzvjyrpn
- Original: https://github.com/simstudioai/sim/releases/tag/v0.5.81

## Features

**Azure Anthropic Integration**
- Added full Azure Anthropic support to enable developers to run Claude models through Azure infrastructure
- Implemented backwards-compatible chat completions API to ensure existing implementations continue working without modification
- Integrated Claude Opus 4.6 model support for latest AI capabilities

**Confluence Integration Expansion**
- Expanded Confluence endpoint coverage to support more workflow automation scenarios with document management systems

## Bug Fixes and Improvements

- **Client Workflow State**: Fixed correct client workflow state override handling to ensure reliable workflow execution
- **Parameter Handling**: Improved canonical parameter handling with enhanced manual validations and streamlined parameter resolution logic
- **Tool Tracing**: Updated tracespans to accurately display inputs for successive identical tool calls, improving debugging visibility

## Developer Impact

These changes enable more flexible AI infrastructure options through Azure integration while maintaining API compatibility. The expanded Confluence support and improved parameter handling make Sim Studio more robust for enterprise workflow automation scenarios.

---

### Linear MCP for product management
- Published: 2026-02-05T16:56:53+00:00
- Products: Linear
- Tags: release, api, feature, integration, deprecation, platform
- URL: https://standarddb.com/linear/linear-expands-mcp-server-with-initiatives-and-milestones-for-claude-cursor--tdbzzvhq
- Original: https://linear.app/changelog/2026-02-05-linear-mcp-for-product-management

## Linear MCP Expansion for Product Management

Linear's MCP server now includes comprehensive support for product management workflows. Product managers can now create and edit initiatives, manage initiative updates, create and edit project milestones, and update project status directly from AI tools like Claude and Cursor. This allows teams to keep Linear synchronized with external workflows without switching contexts.

**New MCP capabilities:**
- Create and edit initiatives and initiative updates
- Create and edit project milestones and project updates
- Manage project labels
- Load images within MCP interactions
- Support for loading Linear resources through URLs

The expansion includes performance improvements through better tool documentation and reduced token usage, making it more efficient for AI-powered interactions.

## Platform Updates

In addition to MCP enhancements, Linear introduces **nested sub-issues** display, allowing teams to visualize issue hierarchies directly in issue lists for better context and progress tracking.

**Team lifecycle management** is now simplified with the ability to retire teams. When teams are no longer active, admins and workspace owners can retire them to reduce sidebar clutter while preserving all project and issue history as read-only archives. Retired teams can be restored at any time.

## Breaking Changes

Linear is **deprecating the `/sse` MCP endpoint** in favor of HTTP streams. Clients using `https://mcp.linear.app/sse` should migrate to `https://mcp.linear.app/mcp`. Deprecation errors will be rolled out gradually over two months. All modern MCP clients already support HTTP streams as the more reliable transport.

## Additional Improvements

The release includes numerous UX and integration improvements: Slack message forwarding now preserves file attachments when creating issues, email intake generates AI-powered titles for messages with missing subjects, iOS app now supports creating initiatives, and notifications alert assignees when issues are deleted. Security enhancements restrict login method bypass to workspace owners only in Enterprise workspaces. The update also fixes critical bugs in the action menu (CJK IME responsiveness), dashboards, editor functionality, and search behavior.

---

### Introducing SyGra Studio
- Published: 2026-02-05T16:52:28+00:00
- Products: 
- Tags: release, feature, platform, open-source
- URL: https://standarddb.com/huggingface/sygra-studio-v2-0-0-debuts-visual-no-code-editor-for-synthetic-data-generation--pwbrttgs
- Original: https://huggingface.co/blog/ServiceNow-AI/sygra-studio

## Overview

SyGra Studio v2.0.0 introduces a unified visual environment for synthetic data generation, replacing manual configuration files with an interactive canvas-based editor. The studio maintains backward compatibility with existing SyGra workflows while simplifying the data generation process through graphical composition.

## Key Features

**Visual Workflow Builder**
- Drag-and-drop node composition for LLM-based pipelines
- Support for multiple LLM providers (OpenAI, Azure OpenAI, Ollama, Vertex AI, Bedrock, vLLM, custom endpoints)
- Auto-discovery of state variables from data sources for prompt templating

**Data Source Integration**
- Direct connectors for Hugging Face datasets, file systems, and ServiceNow instances
- Row preview capability before execution
- Automatic column-to-variable mapping

**Execution & Monitoring**
- Real-time node progress tracking during workflow execution
- Per-run visibility into token costs, latency, and guardrail outcomes
- Inline debugging with logs and breakpoints
- Monaco-backed code editor for advanced customization

**Artifact Generation**
- Visual designs automatically generate production-ready YAML/JSON configuration files
- Stored in standard `tasks/` directory structure for version control
- Execution history preserved in `.executions/` directory

## Developer Experience

The platform maintains API compatibility with existing SyGra workflows—anything built in Studio generates the same configuration format as manual scripts. Users can preview generated artifacts in a code panel before execution, providing transparency and control over the underlying configuration.

## Getting Started

New workflows can be created by selecting data sources, configuring LLM nodes with prompts and output schemas, and chaining multiple generation steps through state variables. The example included demonstrates a code-assistant workflow that ingests the Glaive Code Assistant dataset and iteratively refines outputs through critique feedback loops.

---

### Use Claude Opus 4.6 on AI Gateway
- Published: 2026-02-05T13:00:00+00:00
- Products: Vercel AI SDK, Vercel
- Tags: feature, api, model, integration
- URL: https://standarddb.com/vercel/vercel-ai-gateway-adds-claude-opus-4-6-with-adaptive-thinking-and-1m-context--tsyccqfr
- Original: https://vercel.com/changelog/claude-opus-4.6-on-ai-gateway

## Claude Opus 4.6 Now on AI Gateway

Anthropic's latest flagship model, Claude Opus 4.6, is now available through Vercel's AI Gateway. The model is built to power agents handling real-world work and excels across the entire development lifecycle, from programming and analysis to creative tasks.

### Key Features

**Adaptive Thinking:** Opus 4.6 introduces a new adaptive thinking parameter that lets the model decide when and how much to reason internally. This approach delivers more efficient responses while maintaining quality—providing equal or better performance than extended thinking mode with less computational overhead.

**Extended Context:** This is the first Opus model to support the extended 1M token context window, enabling developers to work with larger codebases, documents, and conversation histories in a single request.

**Interleaved Tool Use:** The model can now interleave thinking and tool calls within a single response, enabling more sophisticated agentic workflows.

### How to Get Started

To use Opus 4.6, set the `model` parameter to `anthropic/claude-opus-4.6`. The adaptive thinking feature is configured via `providerOptions`, with an optional `effort` parameter to control reasoning intensity:

```javascript
const result = streamText({
  model: 'anthropic/claude-opus-4.6',
  prompt: 'Your prompt here',
  providerOptions: {
    anthropic: {
      thinking: { type: 'adaptive' },
      effort: 'max',
    },
  },
});
```

### AI Gateway Benefits

Beyond model access, AI Gateway provides a unified interface for calling models, tracking usage and costs, and configuring retries and failover strategies. It includes built-in observability, Bring Your Own Key (BYOK) support, and intelligent provider routing with automatic retries to ensure higher uptime than direct provider access.

---

### Opus 4.6 is now available in Windsurf
- Published: 2026-02-05T12:00:00+00:00
- Products: 
- Tags: 
- URL: https://standarddb.com/windsurf/opus-4-6-is-now-available-in-windsurf--lqcvntrw
- Original: https://windsurf.com/blog/opus-4.6



---

### Introducing Trusted Access for Cyber
- Published: 2026-02-05T10:00:00+00:00
- Products: OpenAI API, ChatGPT, OpenAI
- Tags: feature, api, security, model, platform
- URL: https://standarddb.com/openai/openai-launches-trusted-access-for-cyber-commits-10m-in-api-credits-for--nlryjynz
- Original: https://openai.com/index/trusted-access-for-cyber

## Introducing Trusted Access for Cyber

OpenAI is launching **Trusted Access for Cyber**, a trust-based framework designed to give cybersecurity defenders priority access to its most capable frontier models while implementing safeguards against misuse. This approach addresses a core challenge: distinguishing between legitimate defensive work (patching vulnerabilities) and malicious activity (exploiting them).

### How It Works

The framework operates on two tracks:

- **Individual users** can verify their identity at chatgpt.com/cyber to access enhanced cyber capabilities
- **Enterprise teams** can request trusted access through their OpenAI representative for organization-wide default access

Security researchers and teams needing even more permissive models for advanced defensive work can apply to an invite-only program. All users with trusted access must comply with OpenAI's Usage Policies and Terms of Use.

### Safety and Safeguards

GPT-5.3-Codex includes built-in mitigations trained to refuse clearly malicious requests like credential theft. The model is paired with automated classifier-based monitors to detect suspicious cyber activity. OpenAI expects to refine these policies and classifiers over time based on feedback from early participants. Prohibited activities include data exfiltration, malware creation or deployment, and unauthorized testing.

### Financial Support for Defenders

OpenAI is committing **$10 million in API credits** through its expanded Cybersecurity Grant Program. The initiative targets teams with proven track records identifying and remediating vulnerabilities in open source software and critical infrastructure. Interested teams can apply through OpenAI's official grants form.

### Context

As frontier models grow more capable of autonomous, long-duration tasks, the potential for both defensive and malicious use increases. OpenAI positions this initiative as critical infrastructure for the broader ecosystem—enabling faster security improvements across software while reducing friction for legitimate cybersecurity work.

---

### Introducing OpenAI Frontier
- Published: 2026-02-05T06:00:00+00:00
- Products: OpenAI API, OpenAI
- Tags: release, feature, platform, api
- URL: https://standarddb.com/openai/openai-launches-frontier-platform-for-enterprise-ai-agent-deployment-and--qjxflrzk
- Original: https://openai.com/index/introducing-openai-frontier

## OpenAI Frontier: Enterprise AI Agent Platform

OpenAI has announced **Frontier**, a new enterprise platform addressing a critical gap in AI agent deployment. While models have become increasingly capable, enterprises struggle to move AI agents from experimental pilots into production across their organizations. Frontier provides an end-to-end solution for building, deploying, and managing AI agents that can perform real work at scale.

### Key Capabilities

The platform is built around four foundational principles:

- **Business Context**: Connects siloed data warehouses, CRM systems, ticketing tools, and internal applications to create a semantic layer that AI coworkers can understand and reference
- **Agent Execution**: Enables agents to reason over data and complete complex tasks—including file operations, code execution, and tool integration—in a dependable, open environment
- **Learning & Improvement**: Agents build memories from past interactions, turning experience into context that improves performance over time
- **Identity & Governance**: Provides agents with clear permissions, boundaries, and access controls that teams can trust

### Architecture & Integration

Frontier is designed to work with existing enterprise infrastructure without requiring platform migrations. The system supports:

- Deployment across local environments, enterprise cloud infrastructure, and OpenAI-hosted runtimes
- Integration with ChatGPT Enterprise, OpenAI's Atlas workflow tool, and existing business applications
- Support for agents developed in-house, acquired from OpenAI, or integrated from third-party vendors
- Low-latency access to OpenAI's models for time-sensitive operations

### Early Adoption

Early adopters include HP, Intuit, Oracle, State Farm, Thermo Fisher, and Uber, with additional pilots underway at companies like BBVA, Cisco, and T-Mobile. Real-world impact cited includes a manufacturer reducing production optimization from six weeks to one day and an energy producer increasing output by up to 5%.

---

### v0.5.80
- Published: 2026-02-05T02:42:13+00:00
- Products: Sim Studio
- Tags: release, feature, api, bugfix, performance
- URL: https://standarddb.com/sim/sim-v0-5-80-adds-async-job-execution-execution-timeouts-and-block-locking--dkkfnwhr
- Original: https://github.com/simstudioai/sim/releases/tag/v0.5.80

## New Features

**Async Job Execution & Timeouts**: The release adds async job queue backends for background task execution, enabling workflows to run asynchronously. Execution timeouts are now configurable, with cloud instances supporting up to 50 minutes and self-hosted deployments maxed out at the same limit.

**UI and Editor Enhancements**: New features include the ability to lock blocks on canvas (preventing accidental edits), note block preview newline support, and added documentation links in the editor for improved developer experience.

**Enterprise Support**: Enterprise modules have been added to support advanced deployment scenarios and organizational needs.

## Improvements & Fixes

**Provider Compatibility**: The OpenAI integration has been migrated to the responses API, and tool calling message formats have been corrected across all LLM providers to ensure consistent behavior.

**Bug Fixes**: Notable fixes address Mistral v2 configuration restoration, nested object serialization in form-urlencoded HTTP bodies, workflow color preservation during import, child workflow response block errors in MCP contexts, and stale execution cleanup integer overflow issues.

**Infrastructure**: Redis client error handling has been improved, file object handling is now consistent across the codebase, and billing actor logging provides better consistency.

## Action Items

- Review timeout configurations if you're running self-hosted deployments
- Test async job execution with your existing workflows to leverage background processing
- Update any custom scripts relying on tool calling message formats across different providers

---

### v0.15.5-rc4: ollamarunner: Fix off by one error with numPredict
- Published: 2026-02-05T01:14:24+00:00
- Products: 
- Tags: release, bugfix
- URL: https://standarddb.com/ollama/ollama-v0-15-5-rc4-fixes-off-by-one-error-in-token-prediction-limits--wmdwxnzx
- Original: https://github.com/ollama/ollama/releases/tag/v0.15.5-rc4

## Off-by-One Error in Token Prediction Fixed

Ollama v0.15.5-rc4 addresses a bug in the `numPredict` parameter that affected token generation limits.

### The Problem

When `numPredict` was set to limit the number of generated tokens:
- Users received **one fewer token than requested**
- Token statistics incorrectly reported the number of tokens returned as the limit itself
- This issue did not occur when `numPredict` was unset

### Root Cause

The bug occurred because the limit was being checked during batch setup rather than during actual token prediction. This caused the current batch to terminate prematurely when the limit was reached.

### The Fix

The fix moves the limit check to occur at the point of actual token prediction, ensuring accurate token counts and proper enforcement of the requested limit.

---

### v0.15.1
- Published: 2026-02-05T01:01:39+00:00
- Products: 
- Tags: release, bugfix, security, performance, model
- URL: https://standarddb.com/vllm/vllm-v0-15-1-ships-security-patches-rtx-blackwell-gpu-fixes-and-4x-faster-torch--thqgqtgx
- Original: https://github.com/vllm-project/vllm/releases/tag/v0.15.1

## Security Updates

vLLM v0.15.1 patches two critical security vulnerabilities in upstream dependencies:

- **CVE-2025-69223**: Updated aiohttp dependency to address security vulnerability
- **CVE-2026-0994**: Updated Protobuf dependency to address security vulnerability

Users running production deployments should upgrade to ensure these fixes are applied.

## Hardware Support & Bug Fixes

**RTX Blackwell GPU Support**: Fixed critical issues preventing NVFP4 MoE (Mixture of Experts) models from loading on RTX Blackwell (SM120) workstation GPUs. Additionally resolved FP8 CUTLASS group GEMM kernel selection, which now properly falls back to Triton kernels on SM120 hardware.

**Model Support**: Added support for Step-3.5-Flash model and fixed loading issues for Qwen3-VL-Reranker and Whisper with FlashAttention2.

## Performance Improvements

**torch.compile Startup**: Fixed a regression that significantly increased cold-start compilation time. Llama3-70B cold-start compilation now completes in ~22 seconds, down from ~88 seconds—a **4x improvement**.

**MoE Optimization**: Optimized the Mixture of Experts forward pass by caching layer name computation, reducing redundant operations.

## Additional Fixes

- Resolved prefix cache hit rate issue with GPT-OSS style hybrid attention models (previously hitting 0%)
- Enabled Triton MoE backend for FP8 per-tensor dynamic quantization
- Fixed speculative decoding metrics crash when no tokens are generated
- Fixed ROCm skinny GEMM dispatch logic
- Pinned LMCache dependency >= v0.3.9 for API compatibility

Developers using vLLM with Blackwell GPUs or large model inference should prioritize this upgrade for both security and performance improvements.

---

### v0.15.2rc0: [Bugfix] Disable TRTLLM attention when KV transfer is enabled (#33192)
- Published: 2026-02-05T00:49:18+00:00
- Products: 
- Tags: release, bugfix
- URL: https://standarddb.com/vllm/vllm-v0-15-2rc0-fixes-trtllm-attention-conflict-with-kv-cache-transfer--lbzsnlgl
- Original: https://github.com/vllm-project/vllm/releases/tag/v0.15.2rc0

## vLLM v0.15.2rc0 Released

vLLM has released v0.15.2rc0, a release candidate focused on stability improvements for production deployments.

### Key Fix

This release includes a critical bugfix that addresses a conflict between two performance optimization features:

- **TRTLLM Attention**: An optimized attention mechanism from TensorRT-LLM
- **KV Cache Transfer**: A feature for efficient key-value cache management

When both features were enabled simultaneously, they could cause incorrect behavior or performance degradation. The fix automatically disables TRTLLM attention when KV cache transfer is active, ensuring safe and predictable operation.

### What Developers Need to Know

If you're using vLLM with either TRTLLM attention or KV cache transfer enabled, upgrading to this release is recommended to avoid potential conflicts. The fix is transparent—no configuration changes are required. The system will automatically select the appropriate attention mechanism based on your settings.

This is a release candidate (rc0), intended for testing before the final v0.15.2 release.

---

### Release: Slack CLI v3.12.0
- Published: 2026-02-05T00:00:00+00:00
- Products: 
- Tags: release, feature, sdk
- URL: https://standarddb.com/slack/slack-cli-v3-12-0-adds-hot-reloading-and-pyproject-toml-support-for-python--grmlndyn
- Original: https://docs.slack.dev/changelog/2026/02/05/slack-cli

## Key Updates

**File Watching and Live Reloading**

The `slack run` command now supports automatic file watching and live reloading for Bolt for JavaScript and Bolt for Python projects. When you modify a file, the development server automatically restarts without manual intervention. Additionally, when the `manifest.json` file changes, the CLI calls the manifest update API (requires `manifest.source: "local"` in `.slack/config.json`).

**Python pyproject.toml Support**

Python projects can now use `pyproject.toml` for dependency management in addition to or instead of `requirements.txt`. The `slack create` and `slack init` commands will automatically detect `pyproject.toml` files and add `slack-cli-hooks<1.0.0>` to the dependency section when missing, with clear installation instructions.

**Updated Command Outputs**

The `slack create` command now displays "AI Agent app" instead of "AI app" to reflect Slack's updated terminology for Agent app features.

**Bug Fixes and Improvements**

- The `slack app settings` command now opens the App Settings home page when run outside a project context, allowing developers to view their complete app list
- Removed support for the deprecated `.slackignore` file, which was unused since the Slack CLI v1.0.0 release
- Added debug logging when manifest changes are detected but not reinstalled due to remote manifest configuration

## Minimum Version Requirements

To use file watching with server restarts, ensure you have these minimum hook versions installed:

- **Python Slack Hooks**: v0.3.0 or higher
- **Node Slack Hooks**: v1.3.0 or higher

The CLI should prompt you to update both the Slack CLI and hooks together.

---

### Developer Update - February 2026
- Published: 2026-02-05T00:00:00+00:00
- Products: Supabase
- Tags: release, feature, api, breaking-change, integration, security
- URL: https://standarddb.com/supabase/supabase-adds-privatelink-ethereum-integration-and-claude-connector--njmgttkw
- Original: https://supabase.com/changelog

## Major Feature Releases

**PrivateLink Support**: Supabase now offers PrivateLink connectivity, allowing developers to connect their databases to AWS resources over private networks without exposing traffic to the public internet. All data flows through AWS VPC Lattice infrastructure, significantly improving security for enterprise deployments.

**Ethereum Blockchain Integration**: A new Infura wrapper enables querying real-time Ethereum blockchain data directly via SQL. Developers can now write standard SQL queries against blockchain data, integrating cryptocurrency infrastructure seamlessly with their Postgres databases.

**Claude Connector**: Supabase became an official Claude connector, enabling AI agents to manage databases through natural language. Users can now instruct Claude to handle database operations directly, combining AI capabilities with database management.

## Developer Tools & Best Practices

The platform introduced 30 Postgres best practice rules across 8 categories designed to improve AI agent-generated code quality. These rules work with Claude Code, Cursor, GitHub Copilot, and other AI coding tools. Additionally, SQL snippets can now be saved locally in Studio and stored in Git, making it easier for teams to version control and share database queries.

## Security Changes & Action Required

**Breaking Change**: `pg_graphql` will be disabled by default on new projects within approximately 3 weeks. This aligns with Supabase's security-first approach of minimizing exposed API surface area by default. Existing projects with zero GraphQL requests will also have the extension disabled. **Action**: Developers relying on GraphQL must manually enable `pg_graphql` via the Database Extensions page or add it to their migrations.

## Additional Improvements

- Edge Functions now support drag-and-drop zip file uploads for easier function migration between projects
- Supabase Assistant provides SQL query performance optimization suggestions in the dashboard
- New integration with TRAE SOLO (ByteDance's AI IDE) for managing database, storage, and auth
- Became an official Tailwind CSS partner



---

### GitHub Actions: Early February 2026 updates
- Published: 2026-02-05T00:00:00+00:00
- Products: GitHub Actions, GitHub
- Tags: release, feature, platform, open-source, security
- URL: https://standarddb.com/github/github-actions-launches-runner-scale-set-client-for-custom-autoscaling-across--gcnhbwbm
- Original: https://github.blog/changelog/2026-02-05-github-actions-early-february-2026-updates

## Runner Scale Set Client Now in Public Preview

GitHub Actions has released a new **runner scale set client** in public preview—a standalone Go-based module that lets you build custom autoscaling solutions for GitHub Actions runners without Kubernetes. This new tool interfaces with the same scale set APIs that power the existing Actions Runner Controller (ARC), but provides flexibility for non-Kubernetes environments.

### Key Capabilities

The client delivers platform-agnostic autoscaling with several important features:

- **Multi-platform support**: Works with containers, virtual machines, and bare metal servers across Windows, Linux, and macOS
- **Full provisioning control**: You define how runners are created, scaled, and destroyed based on your specific infrastructure requirements
- **Native multi-label support**: Assign multiple labels to each scale set for optimizing resources across different build types
- **Agentic workflow support**: Integrates with GitHub Copilot coding agent and other agentic scenarios
- **Real-time telemetry**: Monitor job execution and runner performance with built-in metrics

The client orchestrates API interactions with GitHub's scale set infrastructure while you maintain complete control over the actual runner provisioning. GitHub provides the APIs and client libraries; you're responsible for implementing your infrastructure automation and scaling strategies.

### Relationship to Actions Runner Controller

The runner scale set client is **not a replacement** for Actions Runner Controller, which remains the recommended Kubernetes solution. Instead, it's a new tool for interfacing with the same scale set APIs to build custom autoscaling solutions outside of Kubernetes. Multi-label support will be added to ARC in version 0.14.0, available in March 2026.

The [GitHub Actions Runner Scale Set Client](https://github.com/actions/scaleset) is fully open source. Explore the source code, adapt it to your infrastructure needs, and review the [official documentation](https://docs.github.com/actions/hosting-your-own-runners) for additional details.

## Action Allowlisting Now Available for All Plans

Action allowlisting has been expanded to all GitHub plans—Free, Team, and Enterprise. This security feature lets you define exactly which actions and reusable workflows can run in your repositories, enforcing least privilege principles and protecting against potentially compromised actions. See the [GitHub Actions permissions documentation](https://docs.github.com/repositories/managing-your-repositorys-settings-and-features/enabling-features-for-your-repository/managing-github-actions-settings-for-a-repository#allowing-select-actions-and-reusable-workflows-to-run) for configuration details and best practices.

## New Runner Images in Public Preview

### Windows Server 2025 with Visual Studio 2026

A new Windows runner image with Visual Studio 2026 is now available in public preview. Use `windows-2025-vs2026` in your workflow's `runs-on:` target to test against the latest tooling. This provides a safe validation path before VS 2026 is integrated into the default `windows-2025` image on May 4, 2026.

### macOS 26 Intel for Larger Runners

The macOS 26 Intel runner image is available in public preview for larger runners. Update your `runs-on:` target to `macos-26-large` to access the latest macOS and Xcode tooling for Intel-based workflows. Check the [macOS runner documentation](https://docs.github.com/actions/reference/runners/larger-runners#available-macos-larger-runners-and-labels) for complete software and version details.

---

### GitHub Actions: Self-hosted runner minimum version enforcement extended
- Published: 2026-02-05T00:00:00+00:00
- Products: GitHub Actions
- Tags: release, breaking-change, platform
- URL: https://standarddb.com/github/github-actions-extends-self-hosted-runner-upgrade-deadline-to-march-16-2026--vzdpczcq
- Original: https://github.blog/changelog/2026-02-05-github-actions-self-hosted-runner-minimum-version-enforcement-extended

## Timeline Extended

GitHub has extended the deadline for enforcing minimum version requirements on self-hosted runners. Organizations now have until **March 16, 2026** to upgrade all runners to v2.329.0 or later—a one-week extension from the previously announced deadline.

## Brownout Period and Testing

Prior to full enforcement, GitHub will conduct a **brownout period from February 16 to March 16, 2026**, during which configurations from runners below v2.329.0 will be temporarily blocked. This rollout is designed to help organizations identify and test their upgrade procedures with minimal disruption. The brownout schedule includes multiple windows across different timezones to account for global teams (see schedule table for specific times in your region).

## What's Required

Starting March 16, 2026, GitHub Actions will **permanently block all configurations from runners older than v2.329.0** (released October 15, 2025). After this date, runners must meet the minimum version requirement before registration, and older runners won't be able to connect or self-upgrade after running the configuration script.

## Action Items

To prepare for this change:

1. **Download and update**: Retrieve the latest runner release from the [Actions Runner releases page](https://github.com/actions/runner/releases)
2. **Update infrastructure**: Modify installation scripts, container images, and deployment automation to use v2.329.0 or later before executing `./config.sh`
3. **Redeploy runners**: Recreate any runners built from images or templates containing older versions
4. **Review documentation**: Consult the [GitHub Actions upgrade documentation](https://docs.github.com/actions/hosting-your-own-runners/managing-self-hosted-runners/about-self-hosted-runners#communication-requirements) for detailed upgrade guidance

Organizations running self-hosted runners should prioritize these updates during the brownout period to avoid disruption to their CI/CD workflows.

---

### February 5, 2026
- Published: 2026-02-05T00:00:00+00:00
- Products: 
- Tags: model, feature, performance
- URL: https://standarddb.com/deepgram/deepgram-updates-nova-3-multilingual-model-with-improved-code-switching-accuracy--fvdzkcxz
- Original: https://developers.deepgram.com/changelog/2026/2/5

## Nova-3 Multilingual Model Improvements

Deepgram has released an updated **Nova-3 multilingual model** focused on enhancing real-world multilingual speech recognition. The update delivers **accuracy improvements across all supported languages**, with the most significant gains in **code-switching scenarios** where multiple languages are mixed within a single utterance or conversation.

### Key Improvements

- **Lower Word Error Rate (WER)**: Both batch and streaming inference show improved accuracy across all languages supported by the multilingual model
- **Enhanced code-switching handling**: Reduced word drops when languages are mixed, addressing a common challenge in real-world multilingual applications

### What Developers Need to Know

These improvements are available immediately without requiring any changes to your existing API calls or configuration. The update is backward-compatible, so you can start benefiting from the improved accuracy right away.

For additional details on supported languages and model specifications, see the [Models and Language Overview](https://developers.deepgram.com/docs/models-languages-overview) documentation.

---

### Retry delayed events feature now available
- Published: 2026-02-05T00:00:00+00:00
- Products: 
- Tags: feature, api, platform
- URL: https://standarddb.com/slack/slack-adds-delayed-events-replay-feature-to-recover-from-outages--dnfcrfnv
- Original: https://docs.slack.dev/changelog/2026/02/05/retry-events-feature

## Delayed Events Feature

Slack has introduced a **Delayed Events** feature that enables developers to replay past events that their applications may have missed during service outages or other disruptions.

### What Changed

- Apps can now replay events that occurred during periods when your app was unavailable
- Helps recover from missed notifications due to downtime or connectivity issues
- Integrates directly with the existing Events API

### Key Benefits

- **Reliability**: Ensures your app doesn't lose critical event data during outages
- **Recovery**: Ability to catch up on missed events after resolving connectivity issues
- **Seamless Integration**: Works within the existing Events API framework

### Getting Started

For implementation details and full documentation on how to use the Delayed Events feature, refer to the [Events API documentation](https://docs.slack.dev/apis/events-api/#delayed-events-retry).

---

### Introducing Claude Opus 4.6
- Published: 2026-02-05T00:00:00+00:00
- Products: Claude Code, Claude, Anthropic
- Tags: release, model, feature, api
- URL: https://standarddb.com/anthropic/anthropic-releases-claude-opus-4-6-with-1m-token-context-outperforms-gpt-5-2-by--lyqwjbnp
- Original: https://www.anthropic.com/news/claude-opus-4-6

## Claude Opus 4.6 Now Available

Anthropic has released Claude Opus 4.6, a substantial upgrade to its flagship model with significant improvements in reasoning, coding capabilities, and agentic task execution. The new model is immediately available on claude.ai, via the Claude API (using `claude-opus-4-6`), and across major cloud platforms.

## Key Capabilities and Improvements

**Coding and Agentic Performance:**
- Achieves the highest score on Terminal-Bench 2.0, an evaluation of agentic coding and system tasks
- Improved code review and debugging skills with better ability to catch its own mistakes
- Can operate more reliably in larger codebases
- Sustains agentic tasks for longer without degradation

**Context Window:**
- Features a **1M token context window in beta** — a first for Opus-class models, enabling work with significantly larger documents and codebases

**Extended Thinking and Control:**
- Introduces **adaptive thinking**, where the model contextually determines how much to use extended reasoning
- New **effort controls** give developers granular control over intelligence, speed, and cost tradeoffs
- Developers can dial effort from "high" (default) to "medium" for simpler tasks to reduce latency and cost

## Benchmark Performance

Opus 4.6 demonstrates state-of-the-art performance across multiple evaluations:
- **GDPval-AA**: Outperforms OpenAI's GPT-5.2 by ~144 Elo points and its predecessor (Opus 4.5) by 190 points on economically valuable knowledge work tasks in finance, legal, and other domains
- **Humanity's Last Exam**: Leads all frontier models on this complex multidisciplinary reasoning test
- **BrowseComp**: Performs better than any other model on online information retrieval tasks
- **Safety Profile**: Maintains competitive safety metrics with low rates of misaligned behavior across evaluations

## New Product Features

**Claude Code Enhancements:**
- Agent teams feature allows assembly of collaborative agents to work on tasks together

**API Features:**
- **Compaction** enables the model to summarize its own context, allowing longer-running tasks without hitting token limits
- New context management tools for building extended workflows

**Office Suite Integration:**
- Substantial upgrades to Claude in Excel
- Claude in PowerPoint now available in research preview for enhanced presentation capabilities

## Pricing and Availability

Pricing remains unchanged at **$5 per million input tokens and $25 per million output tokens** across all access channels. Early Access partners including Notion, GitHub, Replit, Asana, and Cognition have reported the model excels at complex multi-step tasks and agentic workflows that previously required significant hand-holding.

---

### Claude Opus 4.6 is now generally available for GitHub Copilot
- Published: 2026-02-05T00:00:00+00:00
- Products: GitHub Copilot
- Tags: release, model, feature
- URL: https://standarddb.com/github/claude-opus-4-6-now-available-in-github-copilot-for-all-plan-tiers--nxjsdgpx
- Original: https://github.blog/changelog/2026-02-05-claude-opus-4-6-is-now-generally-available-for-github-copilot

## Claude Opus 4.6 Now Generally Available in GitHub Copilot

Anthropic's latest Claude Opus 4.6 model is now rolling out to GitHub Copilot users across all plan tiers. The model demonstrates strong performance in agentic coding scenarios, particularly for complex tasks that require multi-step planning and tool calling capabilities.

### Plan Availability

Claude Opus 4.6 is available to:
- **Copilot Pro and Pro+ users**
- **Copilot Business accounts**
- **Copilot Enterprise accounts**

### Where to Access

Users can select Claude Opus 4.6 from the model picker across multiple platforms and interfaces:
- **VS Code**: Chat, Ask, Edit, and Agent modes
- **Visual Studio**: Agent and Ask modes
- **github.com**: Web interface
- **GitHub Mobile**: iOS and Android apps
- **GitHub CLI**: Command-line access
- **GitHub Copilot Coding Agent**: Available for Pro and Pro+ accounts

### Rollout and Enablement

The rollout is gradual—if you don't see Claude Opus 4.6 yet, check back soon. For Copilot Enterprise and Business plan administrators, Claude Opus 4.6 access must be explicitly enabled through the Copilot settings policy configuration.

### Next Steps

Explore all available models in GitHub Copilot through the [supported models documentation](https://docs.github.com/copilot/reference/ai-models/supported-models). Share your feedback in the [GitHub Community](https://github.com/orgs/community/discussions/categories/copilot-conversations).

---

### Pinned comments on GitHub Issues
- Published: 2026-02-05T00:00:00+00:00
- Products: GitHub
- Tags: feature, platform
- URL: https://standarddb.com/github/github-adds-pinned-comments-and-anti-spam-nudges-to-issues--hyrkcxdb
- Original: https://github.blog/changelog/2026-02-05-pinned-comments-on-github-issues

## Pinned Comments for Key Updates

You can now pin individual comments to the top of a GitHub Issue using the comment menu. This feature helps surface critical information such as decisions, next steps, and important updates, allowing anyone viewing the issue to quickly understand what matters most without scrolling through the entire thread.

## Reducing Comment Noise

GitHub has introduced a subtle nudge that encourages contributors to use reactions and subscriptions instead of posting low-quality comments like "+1" or "same here." The system displays a warning in the comment box when it detects intent to post such comments, directing users to more effective engagement methods.

**Benefits:**
- **Reduced clutter** – Fewer unnecessary comments keep issues focused and easier to scan
- **Fewer notifications** – Teams receive less notification spam from low-value comments
- **Better support signals** – Reactions provide clear support metrics without cluttering conversation threads

## What Developers Need to Know

These changes roll out automatically to all GitHub Issues. No configuration is required. Teams using GitHub Issues can immediately start pinning important comments and will see the anti-spam nudges prompt users to engage more constructively. This is particularly valuable for high-traffic repositories with large contributor communities.

---

### Introducing GPT-5.3-Codex
- Published: 2026-02-05T00:00:00+00:00
- Products: OpenAI API, OpenAI
- Tags: release, feature, model, api
- URL: https://standarddb.com/openai/openai-releases-gpt-5-3-codex-with-agentic-coding-capabilities-achieves-new-swe--wykfycpg
- Original: https://openai.com/index/introducing-gpt-5-3-codex

## New Agentic Coding Frontier

OpenAI has introduced **GPT-5.3-Codex**, marking a significant advancement in AI-assisted software development. The model combines the coding capabilities of GPT-5.2-Codex with the reasoning and professional knowledge of GPT-5.2, operating 25% faster while handling complex, long-running tasks. Unlike its predecessors, GPT-5.3-Codex can work autonomously on extended projects while remaining interactive—developers can steer and provide feedback without losing context.

Notably, GPT-5.3-Codex played an instrumental role in its own development. The Codex team used early versions to debug training processes, manage deployments, and diagnose test results, demonstrating the model's ability to accelerate its own improvement cycle.

## Benchmark Performance and Capabilities

GPT-5.3-Codex achieves state-of-the-art performance across multiple benchmarks:

- **SWE-Bench Pro**: Sets new industry high for real-world software engineering tasks across four programming languages, with improved contamination resistance and industry relevance
- **Terminal-Bench 2.0**: Achieves 77.3% accuracy (vs. 64.0% for GPT-5.2-Codex), measuring terminal skills critical for coding agents
- **Token Efficiency**: Delivers stronger performance while consuming fewer tokens, enabling users to build more within usage limits

## Web Development and Long-Running Tasks

GPT-5.3-Codex demonstrates striking capabilities in web development. In testing, the model autonomously iterated on complex games over millions of tokens, building fully functional applications from scratch. It also shows improved intent understanding for day-to-day website development, with better defaults for aesthetic choices, functional layouts, and production-ready designs.

**Developer Action Items**: The model is available via the Codex app waitlist. Developers can now delegate longer-horizon tasks, from multi-day projects to complex tool integration workflows, with improved autonomy and reasoning capabilities.

---

### Opus 4.6, Now Live in Harvey
- Published: 2026-02-05T00:00:00+00:00
- Products: 
- Tags: release, model, integration, api
- URL: https://standarddb.com/harvey/harvey-integrates-anthropic-s-opus-4-6-achieving-90-2-on-legal-reasoning--tyzcxnbg
- Original: https://www.harvey.ai/blog/opus-4-6-now-live-in-harvey

## Claude Opus 4.6 Now Available in Harvey

Harvey has made Claude Opus 4.6, Anthropic's latest frontier model, available to its legal AI platform users. The model builds on Opus 4.5's agentic capabilities with meaningful improvements across legal workflows, particularly in research-intensive work, due diligence, and complex transactional matters.

### Performance Highlights

On Harvey's BigLaw Bench evaluation suite, Opus 4.6 achieved a **90.2% score** — the highest for any Claude model to date. Key performance indicators include:

- **40% of tasks received perfect scores**, demonstrating strong legal reasoning across both litigation and transactional practice areas
- **Deal management, risk assessment, and corporate strategy** tasks were standout performers
- **Superior source grounding** with one complex research query generating over 120 inline citations tied to specific source passages

### Key Improvements

Early access testing revealed several substantive enhancements:

- **Higher substantive accuracy** with improved analytical depth on complex, multi-faceted legal questions
- **Cleaner output formatting** — the model resolves stylistic issues from prior Claude versions, eliminating unnecessary preamble text
- **Thorough source grounding** — citations are meticulously tied to specific passages in source material

### Availability

Opus 4.6 is live for Harvey users now, extending Harvey's existing integration with Anthropic's Claude models. The model is positioned to handle the most challenging legal workflows where precision and source accountability are critical.

---

### Improved pull request Files changed – February 5 updates
- Published: 2026-02-05T00:00:00+00:00
- Products: GitHub
- Tags: release, feature, performance, bugfix, platform
- URL: https://standarddb.com/github/github-improves-pull-request-files-changed-tab-with-codeowners-validation-and--jpjvzqdj
- Original: https://github.blog/changelog/2026-02-05-improved-pull-request-files-changed-february-5-updates

## CODEOWNERS Validation

The new "Files changed" experience now includes `CODEOWNERS` validation, a feature gap from the classic interface. Required reviewers are now correctly surfaced based on CODEOWNERS rules, ensuring review requirements are accurately enforced before merge. This brings parity with the legacy interface while maintaining the benefits of the redesigned experience.

## Performance Optimizations

Performance improvements are the headline of this release, with particular focus on large pull requests:

- **Responsiveness**: Pull request diffs respond up to **67% faster** to clicks, typing, and scrolling
- **Tab Navigation**: Switching between Conversations and Files tabs now takes a few seconds instead of 10+ seconds
- **Memory Usage**: Fixed high memory consumption when reviewing large pull requests
- **Virtualization**: Only the largest pull requests now use virtualization for enhanced scrolling performance
- **File Tree Resizing**: Smoother and more responsive resize interactions, even with many files
- **Safari Performance**: Improved performance and stability on Safari

## Mobile and UI Improvements

The update rolls out multiple fixes for small screen rendering, addressing layout, spacing, and overflow issues to ensure diffs, comments, headers, and controls display correctly on mobile and smaller viewports.

## Bug Fixes

Additional stability improvements include:
- Notifications opened directly to Files changed now correctly mark as read
- Fixed sticky behavior on file diff headers during scrolling
- Improved code snippet styling in thread replies for better contrast
- Fixed errors when commenting across large index differences in diffs

## Developer Action Items

Developers and teams should notice immediate improvements when reviewing pull requests, particularly for large diffs. The CODEOWNERS validation ensures review workflows function correctly without manual enforcement.

---

### v0.15.5-rc3: qwen3next: fix issue in delta net (#14075)
- Published: 2026-02-04T21:40:38+00:00
- Products: 
- Tags: bugfix, release
- URL: https://standarddb.com/ollama/ollama-v0-15-5-rc3-fixes-qwen3-delta-net-tensor-broadcasting-issue--msjdkjvm
- Original: https://github.com/ollama/ollama/releases/tag/v0.15.5-rc3

## Qwen3 Delta Network Broadcasting Fix

Ollama v0.15.5-rc3 addresses a bug in the Qwen3 model implementation affecting the delta network layer's backward pass computation.

### The Issue

The `gDiffExp` gradient tensor was being broadcast across the wrong axis during multiplication with the key tensor (k), causing incorrect gradient computation in the model's attention mechanism.

### The Fix

The fix reshapes `gDiffExp` to the correct dimensions of `[1, chunkSize, nChunks, ...]` to ensure proper tensor alignment and broadcasting during the multiplication operation. This ensures gradients are computed correctly during backpropagation.

### Impact

This bugfix is particularly important for users training or fine-tuning Qwen3 models with Ollama, as incorrect gradient computation would lead to degraded model performance and unreliable training results.

---

### Build with Kimi K2.5 Multimodal VLM Using NVIDIA GPU-Accelerated Endpoints 
- Published: 2026-02-04T19:46:33+00:00
- Products: 
- Tags: feature, model, api, integration, platform
- URL: https://standarddb.com/nvidia/kimi-k2-5-multimodal-vlm-now-available-on-nvidia-gpu-accelerated-endpoints-with--qhbylcyr
- Original: https://developer.nvidia.com/blog/build-with-kimi-k2-5-multimodal-vlm-using-nvidia-gpu-accelerated-endpoints

## Kimi K2.5 Model Architecture

Kimi K2.5 is a general-purpose multimodal vision language model trained using the open-source Megatron-LM framework. The model contains **1 trillion total parameters** with a 3.2% activation rate per token, achieved through a mixture-of-experts (MoE) architecture with 384 experts and 1 shared dense layer. This design enables efficient processing across multiple modalities with only 32.86B active parameters per token.

The model supports **text, image, and video inputs** across a 262K token context window, making it well-suited for agentic AI workflows, reasoning, coding, mathematics, and chat applications. Kimi developed the **MoonViT3d Vision Tower** to handle visual processing, converting images and video frames into embeddings that integrate seamlessly with the language model component.

## Free Prototyping and Deployment Options

Developers can access Kimi K2.5 for free prototyping through **NVIDIA GPU-accelerated endpoints** on [build.nvidia.com](http://build.nvidia.com/) as part of the NVIDIA Developer Program. The browser-based experience allows testing with custom data without requiring API setup.

**API Integration**: The model is accessible through an OpenAI-compatible API endpoint at `https://integrate.api.nvidia.com/v1/chat/completions` with support for tool calling via the `tools` parameter. NVIDIA NIM microservices containers for production inference are coming soon.

**Deployment with vLLM**: For teams deploying with the vLLM serving framework, detailed instructions and recipes are available in the official vLLM documentation. Additional deployment and fine-tuning options are available through the NVIDIA NeMo Framework for domain-specific customization.

## Getting Started

Developers can start building immediately by visiting the Kimi K2.5 page on build.nvidia.com with free registration in the NVIDIA Developer Program. The model supports standard chat completion parameters including temperature, top_p, frequency penalties, and streaming responses.

---

### Showing tool calls and other improvements to Copilot chat on the web
- Published: 2026-02-04T18:17:01+00:00
- Products: GitHub Copilot
- Tags: feature, platform
- URL: https://standarddb.com/github/github-copilot-chat-adds-visible-tool-calls-and-chat-export-to-web-interface--ckrgqgvz
- Original: https://github.blog/changelog/2026-02-04-showing-tool-calls-and-other-improvements-to-copilot-chat-on-the-web

## Tool Calls Visibility

You'll now see clear tool call details in real-time when you send Copilot a message. This transparency helps you understand the steps Copilot is taking to fulfill your request, course-correct if something fails, and see references generated by each tool call. This makes the AI's reasoning process more interpretable for developers.

## Chat Export Functionality

Copilot Chat conversations can now be easily exported to either JSON or Markdown formats. To export a conversation, select the chat you want to download, press the "Manage chat" menu, and choose **Export**. This allows developers to reference conversations later, archive important discussions, or share context with team members.

## Improved Repository Attachments

The repository attachment experience has been enhanced with better repository recommendations when searching. Users can now find the right repository faster, streamlining the process of providing Copilot with codebase context for more targeted assistance.

These improvements focus on making Copilot Chat feel smoother and faster on github.com, addressing common developer workflows around transparency, portability, and context setup.

---

### GitHub Copilot in Visual Studio Code v1.109 – January Release
- Published: 2026-02-04T18:00:01+00:00
- Products: GitHub Copilot
- Tags: release, feature, api, integration, security, platform
- URL: https://standarddb.com/github/github-copilot-v1-109-adds-claude-agent-support-and-multi-agent-orchestration--xqwljjwx
- Original: https://github.blog/changelog/2026-02-04-github-copilot-in-visual-studio-code-v1-109-january-release

## Agent Extensibility & Claude Integration

GitHub Copilot v1.109 now offers **public preview support for Claude agents** powered by Anthropic's official Claude Agent SDK. This integration allows developers to delegate complex tasks directly from their GitHub Copilot subscription using Claude models, seamlessly within Visual Studio Code. Additionally, richer chat interactions with MCP (Model Context Protocol) apps enable more tool-driven, interactive experiences.

## Enhanced Chat & Reasoning

The release brings significant UX improvements to Copilot Chat:

- **Faster streaming responsiveness** for snappier, more responsive conversations
- **Higher-quality reasoning results** using thinking tokens to improve reliability and task execution
- **Better visibility into agent reasoning** with agents asking clarifying questions instead of making assumptions
- **Revamped inline editor chat** for more natural, in-context coding assistance

## Multi-Agent Orchestration & Session Management

Developers can now seamlessly coordinate work across multiple agent types:

- **Unified agent management** across Copilot, Claude, local agents, background agents, and cloud agents
- **Parallel subagent execution** for faster task completion
- **Improved session visibility** with status indicators showing which sessions need attention
- **Seamless handoff** allowing developers to jump into agent work without disrupting momentum

## Agent Customization & Intelligence

New capabilities for tailoring agent behavior to team needs:

- **Agent orchestrations** for building repeatable, team-specific workflows
- **Agent Skills** combined with org-wide customization ensure consistent outcomes across developers
- **Copilot Memory** helps agents retain relevant context across interactions for smarter task handling
- **Faster code search** via external indexing improves retrieval speed in large repositories

## Security & Control Improvements

The release prioritizes safety in agent-driven development:

- **Terminal command sandboxing** (experimental, macOS/Linux) increases confidence when agents propose or execute commands
- **Effective auto-approval rules** improve safety and control while reducing unnecessary user prompts
- **Enhanced terminal lifecycle management** with timeout, await, and kill capabilities for better control over background commands

## Additional VS Code Updates

Beyond Copilot enhancements, this release includes general VS Code productivity improvements: integrated browser for testing apps without leaving the editor, terminal quality-of-life enhancements, coding improvements, and new extension APIs.

---

### Nemotron ColEmbed V2: Raising the Bar for Multimodal Retrieval with ViDoRe V3’s Top Model
- Published: 2026-02-04T15:00:40+00:00
- Products: 
- Tags: release, model, feature
- URL: https://standarddb.com/huggingface/nvidia-releases-nemotron-colembed-v2-multimodal-models-8b-variant-ranks-1-on--dtxwfmgd
- Original: https://huggingface.co/blog/nvidia/nemotron-colembed-v2

## New Multimodal Embedding Models

NVIDIA introduces the Nemotron ColEmbed V2 family, a set of late-interaction embedding models available in three sizes—3B, 4B, and 8B parameters. These models are specifically designed for accurate multimodal retrieval across heterogeneous documents containing text, images, tables, charts, and other visual components.

## Architecture and Approach

Unlike single-vector embedding approaches, Nemotron ColEmbed V2 adopts a **late-interaction architecture** inspired by ColBERT. This enables fine-grained token-level interactions between queries and documents:

- Each query token embedding interacts with all document token embeddings using the MaxSim operator
- The operator selects maximum similarity for each query token and sums them for a final relevance score
- Models use bi-directional self-attention (instead of causal) for richer representation learning
- Supports both textual and visual token interactions

## Performance and Benchmarks

The models achieve state-of-the-art results on the ViDoRe V3 benchmark for visual document retrieval:

- **nemotron-colembed-vl-8b-v2**: Ranks #1 with 63.42 NDCG@10 (8.8B parameters)
- **nemotron-colembed-vl-4b-v2**: Ranks #3 with 61.54 NDCG@10 (4.8B parameters)
- **llama-nemotron-colembed-vl-3b-v2**: Ranks #6 with 59.79 NDCG@10 (4.4B parameters)

## Use Cases and Distinction

These models are intended for researchers and enterprises prioritizing accuracy in visual document retrieval applications, distinguishing them from NVIDIA's earlier 1B single-vector model optimized for efficiency and storage. Key applications include:

- Multimodal RAG systems with textual queries retrieving document images
- Multimedia search engines and cross-modal retrieval
- Conversational AI with rich input understanding
- Enterprise document processing systems

All three models are now available on Hugging Face.

---

### Parallel's Web Search and tools are live on Vercel
- Published: 2026-02-04T13:00:00+00:00
- Products: Vercel AI SDK, Vercel
- Tags: feature, integration, api, model
- URL: https://standarddb.com/vercel/vercel-integrates-parallel-web-search-across-ai-gateway-and-ai-sdk--zzkkngfb
- Original: https://vercel.com/changelog/parallel-web-search-is-now-on-ai-gateway

## Parallel Web Search Integration

Vercel has launched Parallel's web search and extraction tools across its platform, making real-time information retrieval available to all developers building AI applications. Unlike provider-specific web search solutions that lock you into particular models, Parallel's tool works universally across all LLM providers through Vercel's AI Gateway.

## How It Works

**Universal Compatibility**: Simply configure `parallel_search: gateway.tools.parallelSearch()` in your tools configuration to enable web search for any model. The tool automatically extracts relevant excerpts from web pages, optimized for agentic tasks and multi-step reasoning workflows.

**Flexible Configuration Options**:
- Set `mode: 'one-shot'` for single queries or `mode: 'agentic'` for multi-step reasoning with token-efficient results
- Control cache freshness with `maxAgeSeconds` for time-sensitive queries
- Restrict results to trusted domains with `sourcePolicy` including domain whitelists and date filters
- Configure result limits to control costs and token usage

## Integration Paths

**AI Gateway**: Use directly in any application with streamlined configuration and support for all major LLM providers.

**AI SDK**: Install the `@parallel-web/ai-sdk-tools` package for native integration in Node.js/JavaScript projects.

**Vercel Marketplace**: Access all Parallel products (Search, Extract, Task, Findall, and Monitoring) with single sign-on and centralized billing through your Vercel account.

## Pricing and Getting Started

Parallel web search requests are charged at $5 per 1,000 requests (including up to 10 results), with additional results beyond 10 costing $1 per 1,000. Developers can start for free or contact Vercel's sales team for Pro or Enterprise pricing. A Next.js starter template is available to quickly integrate Parallel's web research APIs.

---

### Build logs now support interactive links
- Published: 2026-02-04T13:00:00+00:00
- Products: Vercel
- Tags: feature, release, platform
- URL: https://standarddb.com/vercel/vercel-build-logs-now-support-clickable-links-for-faster-debugging--bnbnhlqk
- Original: https://vercel.com/changelog/build-logs-now-support-interactive-links

## Interactive Links in Build Logs

Vercel has enhanced its build logs interface by making URLs clickable and interactive. Previously, developers had to manually copy and paste links from build logs to investigate issues or access relevant documentation. This update eliminates that friction.

### What Changed

- **Direct navigation**: URLs in build logs are now clickable links
- **External link handling**: External links automatically open in new browser tabs
- **Seamless workflow**: No more copying and pasting URLs during debugging sessions

### Benefits

This improvement streamlines the developer experience when troubleshooting build failures and accessing related resources. When a build fails with error messages or links to documentation, developers can now click directly to those resources, saving time and reducing context switching.

### Getting Started

No configuration or action is required. This feature is automatically available for all Vercel deployments. Learn more about accessing build logs in the [official documentation](https://vercel.com/docs/deployments/logs).

---

### Now in Nature: Synthesizing scientific literature with retrieval-augmented LMs
- Published: 2026-02-04T08:00:00+00:00
- Products: 
- Tags: model, release, open-source, api
- URL: https://standarddb.com/ai2/ai2-publishes-nature-paper-on-retrieval-augmented-model-for-scientific--prqfxdnb
- Original: https://allenai.org/blog/openscholar-nature

## Reliable Scientific Synthesis with Verifiable Citations

Keeping up with scientific publishing is challenging for researchers, and general-purpose AI systems have struggled with a critical requirement: reliable grounding in peer-reviewed literature. While language models can generate plausible summaries, they often cite irrelevant work or fabricate sources entirely—a problem known as hallucination that undermines trust in AI-assisted research.

## OpenScholar: Building on Retrieval-Augmented Generation

AI2 and University of Washington researchers addressed this challenge by developing **OpenScholar**, an open-source model specifically designed for scientific literature synthesis. The system combines a model trained on scientific synthesis tasks with retrieval-augmented generation (RAG), allowing it to:

- Search a corpus of 45 million open-access scientific papers
- Incorporate relevant papers—including recent publications—into its responses
- Provide verifiable citations for claims made in synthesized literature reviews
- Leverage a full-text snippet index made available through the Semantic Scholar API

## Supporting Resources and Benchmarking

To support evaluation of scientific synthesis systems, the researchers created **ScholarQABench**, the first large-scale multi-domain benchmark for evaluating citation quality and scientific answer generation. The computer science portion (ScholarQA-CS) evolved into ScholarQA-CS2, which is now part of AstaBench.

All research artifacts—including model checkpoints, the retrieval index, training data, and a public demo—are freely available. This research has already influenced downstream work, including the ScholarQA feature now in Asta and the expanded multi-step search capabilities in Deep Research Tulu (DR Tulu).

## What This Means for Research Workflows

By showing that careful retrieval, ranking, and citation handling can substantially improve trustworthiness in scientific contexts, this work provides a foundation for AI systems that "show their work" rather than simply sounding convincing. Researchers can now inspect, validate, and extend these open-source tools for their own domains.

---

### AI Crawl Control - New reference documentation
- Published: 2026-02-04T00:00:00+00:00
- Products: Cloudflare Workers, Cloudflare
- Tags: feature, api, documentation, platform
- URL: https://standarddb.com/cloudflare/cloudflare-releases-ai-crawl-control-reference-documentation-with-graphql-api--fwvtgnxy
- Original: https://developers.cloudflare.com/changelog/2026-02-09-reference-documentation

## AI Crawl Control Reference Documentation Now Available

Cloudflare has released new reference documentation for **AI Crawl Control**, making it easier for developers to manage and monetize access to their content by AI crawlers.

### Key Documentation Resources

The release includes three major documentation components:

- **GraphQL API Reference** — Developers can now query detailed crawler request data, including top paths, referral traffic, and data transfer metrics. The API supports key filters for detection IDs, user agents, and referrer domains.
- **Bot Reference** — Complete detection identifiers and user agents for major AI crawlers from OpenAI, Anthropic, Google, Meta, and other companies.
- **Worker Templates** — Ready-to-deploy Cloudflare Worker templates, including an x402 Payment-Gated Proxy for monetizing crawler access or charging bots while allowing free access to human users.

### What Developers Need to Know

This documentation enables developers to:
- Implement fine-grained control over which crawlers can access their content
- Monitor and analyze crawler traffic patterns
- Monetize crawler access with payment-gated proxies
- Differentiate between AI bots and human users with accurate detection

The reference materials are now available in the [Cloudflare Developers documentation](https://developers.cloudflare.com/ai-crawl-control/).

---

### Choose Your Edit Prediction Provider
- Published: 2026-02-04T00:00:00+00:00
- Products: 
- Tags: feature, api, integration, platform
- URL: https://standarddb.com/zed/zed-adds-pluggable-edit-prediction-with-support-for-six-providers-including--vxjbyzvh
- Original: https://zed.dev/blog/edit-prediction-providers

## Multiple Edit Prediction Providers Now Available

Zed has introduced a pluggable edit prediction provider architecture, giving developers the ability to choose between multiple code prediction services. The supported providers are:

- **Zeta** (default) — Zed's proprietary model trained specifically for edit prediction
- **Mercury Coder** — Uses diffusion-based architecture for parallel output refinement
- **Sweep** — Fast predictions trained with online RL, delivering suggestions in under 100ms
- **Ollama** — Run open-weight models locally (Qwen, CodeLlama, DeepSeek) without sending code off-machine
- **GitHub Copilot** — With Next Edit Suggestions
- **Codestral** — Mistral's code model

## Simplified Provider Architecture

The refactoring moved most edit prediction logic (state management, UI integration, debouncing, caching) into Zed's core, significantly reducing the effort required to add new providers. Developers implementing new providers only need to handle model-specific tasks: prompt construction, API calls, and response parsing. This opens the door for community contributions via pull requests.

## Key Capabilities and Trade-offs

Edit predictions work best when they match individual coding patterns—anticipating edits, suggesting closing braces, or removing unnecessary code. Each provider offers different latency characteristics and performance depending on language, project size, and editing style. Mercury Coder's 1-month free trial is available with a provided API key.

## Getting Started

Developers can configure their preferred provider directly in Zed's Settings Editor and experiment to find the best fit for their workflow. The local Ollama option is particularly valuable for teams with code privacy requirements.



---

### Workflows, Workers - Visualize your Workflows in the Cloudflare dashboard
- Published: 2026-02-04T00:00:00+00:00
- Products: Cloudflare Workers, Cloudflare
- Tags: feature, platform
- URL: https://standarddb.com/cloudflare/cloudflare-workflows-gains-visual-diagram-tool-for-debugging-and-code--mpcrktqj
- Original: https://developers.cloudflare.com/changelog/2026-02-03-workflows-visualizer

## Visual Workflow Diagrams Now Available

Cloudflare has introduced automatic visual diagram generation for Workflows, providing developers with a graphical representation of their workflow code. This new capability parses your Workflow code and generates a visual map showing how steps connect, execute, and interact with one another.

## Key Capabilities

The diagram visualization enables developers to:

- **Understand step connections** – See how each step flows into the next
- **Visualize complex logic** – View loops and nested logic in a clear graphical format
- **Follow conditional paths** – Trace branching logic with ease
- **Collapse and expand** – Toggle between high-level flow views and detailed step-by-step views

## Availability and Access

The feature is currently available in beta for all JavaScript and TypeScript Workflows. Developers can access workflow diagrams directly in the [Cloudflare dashboard](https://dash.cloudflare.com/?to=/:account/workers/workflows) by navigating to their Workflows section. This makes debugging, documentation, and understanding existing workflow logic significantly easier without leaving the dashboard.

---

### GitHub Copilot in Visual Studio — January update
- Published: 2026-02-04T00:00:00+00:00
- Products: GitHub Copilot
- Tags: feature, release
- URL: https://standarddb.com/github/github-copilot-in-visual-studio-gains-syntax-highlighted-completions-and--rvxjytjh
- Original: https://github.blog/changelog/2026-02-04-github-copilot-in-visual-studio-january-update

## Key Features

GitHub Copilot in Visual Studio 2026 introduces several user-requested enhancements designed to improve the coding experience:

**Colorized Code Completions**: Code suggestions now display syntax highlighting, making it easier to visually distinguish variables, functions, keywords, and other language elements. Developers can enable this feature in settings by selecting **Use colorized text for code completions**.

**Partial Acceptance of Completions**: A new click-to-accept feature allows developers to accept only a portion of a code suggestion. Hovering over a suggestion highlights individual segments, and clicking positions the cursor to accept code up to that point, providing more granular control over what gets inserted.

## Additional Improvements

Beyond completions, the update delivers several productivity enhancements:

- **Faster scrolling** for smoother editor navigation
- **HTML-rich copy/paste** support for better content handling
- **Syntactic line compression** for more compact code display

**Streamlined Markdown Preview**: Developers can now switch between Split Preview, Open Preview, and Edit Markdown modes. Zoom controls are now available for Mermaid diagrams, and Copilot Chat includes a Preview button to render and edit generated markdown directly without leaving the editor.

## Getting Started

The January features are available now in Visual Studio 2026. Download the latest version to access all new Copilot capabilities, and visit the [Visual Studio blog](https://devblogs.microsoft.com/visualstudio) for additional details and a full feature roadmap.

---

### Claude and Codex are now available in public preview on GitHub
- Published: 2026-02-04T00:00:00+00:00
- Products: GitHub Copilot, GitHub
- Tags: feature, release, integration, platform
- URL: https://standarddb.com/github/github-copilot-expands-with-claude-and-codex-agents-in-public-preview--zpxyrjww
- Original: https://github.blog/changelog/2026-02-04-claude-and-codex-are-now-available-in-public-preview-on-github

## New Agents Now Available

GitHub has brought Claude (Anthropic) and OpenAI Codex into its Copilot platform as coding agents, available in public preview for Pro+ and Enterprise subscribers. These agents complement the existing Copilot agent and provide developers with additional AI model choices for tackling coding tasks.

## How to Enable and Use

**Setup is straightforward:**
- **Pro+ subscribers** can enable Claude and/or Codex from their Copilot coding agent settings
- **Enterprise admins** configure access at the enterprise and organization levels under AI Controls and Copilot settings
- No additional subscriptions required—access is included with existing Copilot plans

**Multiple ways to work with agents:**
- Create sessions from the **Agents tab** in any enabled repository on GitHub.com or mobile
- Assign issues directly to Claude, Codex, or Copilot using the assignees dropdown
- Request agent help in pull requests with `@claude` or `@codex` mentions
- Start interactive or autonomous sessions in **VS Code 1.109+** via the Agent sessions view

## Integration Points

Agents are accessible across the GitHub ecosystem. Users can start work on the web, continue in VS Code, and track progress in real-time session views. Agents automatically generate draft pull requests for assigned work, iterate on feedback, and provide detailed activity logs for review. Each coding agent session consumes one premium request during the public preview period.

Developers should check the [third-party agents documentation](https://docs.github.com/copilot/concepts/agents/about-third-party-agents) for comprehensive guidance on maximizing these tools.

---

### Discord Patch Notes: February 4, 2026
- Published: 2026-02-04T00:00:00+00:00
- Products: 
- Tags: release, feature, performance, bugfix, api
- URL: https://standarddb.com/discord/discord-ships-rust-audio-backend-for-80-of-traffic-adds-stream-zoom-and-linux--xdvqhblp
- Original: https://discord.com/blog/discord-patch-notes-february-4-2026

## Performance & Infrastructure

Discord has made substantial progress migrating its Audio/Video backend to Rust, with over 80% of traffic now served by the new implementation. Additionally, the team shipped a major desktop rendering optimization that significantly reduced navigation and interaction delays—particularly benefiting users on lower-end hardware. The improvement came from addressing slow CSS selectors rather than backend endpoints or component inefficiencies.

## New Features

**Screenshare & Stream Improvements:** Users can now zoom and pan on screenshares and game streams using mouse scroll wheel or trackpad, addressing a long-requested feature. Stream preview load times have also been optimized in the start stream flow.

**Linux Timestamp Support:** A new `@time` command on desktop automatically generates timezone-aware timestamps, eliminating the need for external tools. The timestamp adapts to each viewer's local timezone.

**Bypass Slowmode Permission:** Server admins can now assign a "Bypass Slowmode" permission to trusted community members. Role integration UI will be available until February 23rd.

## Platform-Specific Fixes

- **Android:** Group DM notifications now display the group name instead of the sender's name (matching iOS behavior). Fixed playback of videos with multiple audio tracks.
- **iOS:** Resolved swipe-scrolling issue on Client Themes list that scrolled at excessive speed. Fixed GIF avatar selection converting to static images.
- **Desktop:** Press ESC while editing your profile to stay in the modal instead of closing it. New role mention click-through feature (100% rolled out) shows users with that role.

## General Bug Fixes & Polish

The patch includes 50+ bug fixes addressing UI alignment, modal rendering, keyboard input (F10+ key persistence), emoji/sticker editing, and various platform-specific crashes. Notable fixes include resolved clipboard clearing, Browse Channels column alignment, member menu styling, and connection settings alignment across platforms.

Vibing Wumpus has been updated to 2.0 (summon with Ctrl+Alt+Shift+W on desktop).

---

### Using Your Design System
- Published: 2026-02-04T00:00:00+00:00
- Products: Magic Patterns
- Tags: feature, platform, release
- URL: https://standarddb.com/magic-patterns/magic-patterns-launches-unified-design-systems-for-consistent-ai-generated-ui--ksljwpjl
- Original: https://www.magicpatterns.com/blog/introducing-design-systems

## Unified Design System Management

Magic Patterns has introduced **Design Systems**, a major consolidation of its previously separate Presets and Libraries features. This new unified approach provides teams with a single source of truth for managing brand identity, component libraries, and design consistency across all AI-generated designs.

## Key Features

**Style Context Files**: Define your brand colors, typography, spacing guidelines, and design principles once. These settings automatically apply to every design generated within that Design System, eliminating repetitive prompt instructions and ensuring team-wide alignment.

**Unified Component Editor**: Manage all components in one centralized location. When you update a component, every design using that Design System instantly gets access to the latest version, ensuring consistency across all outputs.

**Multiple Import Methods**:
- **Screenshots**: Upload images from Figma, production apps, PDFs, or sketches; Magic Patterns generates matching code
- **Figma Integration**: Connect your Figma account and paste frame URLs to import designs with full fidelity, including colors, typography, spacing, and component structure
- **Storybook & Web**: Use the Chrome Extension to import components from Storybook, localhost, or any authenticated page

## Existing User Migration

All existing Presets and Libraries have been automatically migrated to Design Systems. If you had a Preset connected to a Library, both are now merged into a single Design System. Separate Presets or Libraries were converted into their own Design Systems and can be used independently or merged manually.

## Availability

Design Systems is now available for all users. Start creating your first Design System from the dashboard and refer to the [documentation](https://www.magicpatterns.com/docs/documentation/design-systems/overview) for implementation details and best practices.

---

### Queues - Cloudflare Queues now available on Workers Free plan
- Published: 2026-02-04T00:00:00+00:00
- Products: Cloudflare Workers, Cloudflare
- Tags: feature, pricing, release
- URL: https://standarddb.com/cloudflare/cloudflare-queues-launches-on-workers-free-plan-with-10-000-daily-operations--ktlxyvcj
- Original: https://developers.cloudflare.com/changelog/2026-02-04-queues-free-plan

## Cloudflare Queues on Free Plan

Cloudflare has expanded access to **Cloudflare Queues** by adding it to the Workers free plan. This makes reliable message queuing available to developers at no cost, significantly lowering the barrier to entry for building asynchronous, event-driven applications on Cloudflare's edge platform.

### What's Included

Free-tier accounts now get:
- **10,000 operations per day** (combined reads, writes, and deletes)
- Support for up to **10,000 queues** per account
- Message delivery to Cloudflare Workers or HTTP pull consumers
- Unlimited event subscriptions
- All existing Queues features

### Key Limitation

The main trade-off on the free plan is a **24-hour message retention window** versus the 14-day retention available on paid tiers. This is suitable for most real-time, transient workloads but may not work for use cases requiring longer message persistence.

### Getting Started

Developers can start building with Queues immediately using the [official quickstart guide](https://developers.cloudflare.com/queues/get-started/) or explore [pre-built tutorials](https://developers.cloudflare.com/queues/tutorials/) to see common patterns and use cases.

---

### Community Evals: Because we're done trusting black-box leaderboards over the community
- Published: 2026-02-04T00:00:00+00:00
- Products: 
- Tags: feature, platform, api, open-source
- URL: https://standarddb.com/huggingface/hugging-face-launches-community-driven-model-evaluations-with-decentralized--lrlxgvfx
- Original: https://huggingface.co/blog/community-evals

## Decentralized Model Evaluation Now Live

Hugging Face is rolling out a new evaluation framework that addresses the fragmentation and lack of transparency in model benchmarking. Rather than relying on siloed leaderboards or competing claims across papers and platforms, the Hub now enables community members to submit evaluation results transparently.

**How It Works**

- **For Benchmarks:** Dataset repositories can register as benchmarks (MMLU-Pro, GPQA, HLE are initial partners) and automatically aggregate reported results into leaderboards. Each benchmark defines its evaluation spec via `eval.yaml` using the Inspect AI format, making results reproducible.

- **For Models:** Evaluation results are stored in `.eval_results/*.yaml` within model repositories, visible both on model cards and aggregated into benchmark leaderboards. Both official model author results and community pull requests are included.

- **For the Community:** Any user can submit evaluation results via pull request for any model. Results appear immediately as "community" contributions while awaiting review, and users can link to supporting sources like papers, logs, or third-party platforms.

## Why This Matters

The current benchmarking landscape has two critical problems: established benchmarks like MMLU (91%+ saturation) no longer differentiate models effectively, yet reported scores vary across sources. This creates a gap between benchmark numbers and real-world performance.

By exposing evaluation results through a transparent, git-based system with full history and attribution, the Hub addresses score fragmentation and enables the community to build curated leaderboards and dashboards via public APIs. The framework emphasizes reproducibility through open eval specs while acknowledging it won't solve benchmark saturation or prevent test set overfitting—but it makes evaluation practices visible.

## Getting Started

Users can publish evals as YAML files in any model repository's `.eval_results/` directory, browse existing scores on benchmark dataset pages, or register new benchmarks by adding `eval.yaml` and contacting the team. The feature is currently in beta.

---

### Introducing Granola MCP
- Published: 2026-02-04T00:00:00+00:00
- Products: 
- Tags: feature, api, integration, platform
- URL: https://standarddb.com/granola/granola-launches-mcp-integration-bringing-meeting-context-to-claude-chatgpt-and--tysrdbss
- Original: https://www.granola.ai/blog/granola-mcp

## What's New

Granola has launched **Granola MCP**, a Model Context Protocol integration that connects meeting notes to popular AI tools. The integration lets Claude, ChatGPT, Cursor, and other MCP-compatible applications access your Granola meeting context directly within their interfaces.

## Key Use Cases

- **Code generation:** Pull meeting context into Claude Code or Cursor to scaffold features or create tickets based on discussed requirements
- **Sprint planning:** Ask Claude to update Linear boards directly from standup notes
- **Sales operations:** Generate CRM entries and proposals from discovery conversations
- **Documentation:** Use meeting discussions as context for drafting and sharing notes without manual copying

## Getting Started

**Claude and ChatGPT:** Navigate to Settings > Connectors (or Apps), search for Granola, authenticate, and enable when chatting.

**Cursor, Claude Code, or other MCP clients:** Manually configure using `https://mcp.granola.ai/mcp`

Users must be on a **paid plan** to access the feature. Enterprise customers can enable MCP in Settings > Security; additional enterprise features (access controls, usage monitoring, rate limits) are in development.

## Why It Matters

This solves a common friction point: switching between applications to reference meeting context. By bringing meeting data into the tools developers already use, Granola makes it easier to extract actionable insights from conversations without breaking workflow.

---

### H Company's new Holo2 model takes the lead in UI Localization
- Published: 2026-02-03T17:40:14+00:00
- Products: 
- Tags: release, model, feature
- URL: https://standarddb.com/huggingface/h-company-releases-holo2-235b-a22b-reaching-78-5-accuracy-on-ui-localization--rbcgvtlq
- Original: https://huggingface.co/blog/Hcompany/introducing-holo2-235b-a22b

## Holo2-235B-A22B: New SOTA in UI Element Localization

H Company has released **Holo2-235B-A22B Preview**, their latest and largest UI localization model. Available on Hugging Face, this research release achieves record-breaking performance on major benchmarks:

- **78.5% accuracy** on Screenspot-Pro (3-step agentic mode)
- **79.0% accuracy** on OSWorld G
- **70.6% accuracy** on Screenspot-Pro (single-step baseline)

## Agentic Localization for High-Resolution Interfaces

The model introduces a key innovation: **agentic localization**, which allows iterative refinement of UI element predictions. High-resolution 4K interfaces present a challenge—small UI elements are difficult to pinpoint on large displays. By enabling the model to refine its predictions across multiple steps, Holo2 achieves **10-20% relative accuracy improvements** compared to single-pass predictions.

## Infrastructure and Deployment

H Company trained Holo2 models at scale using **SkyPilot**, a unified interface for launching training jobs across multiple cloud providers and Kubernetes clusters. This abstraction simplifies infrastructure management, allowing researchers to focus on model development rather than maintaining deployment configurations.

## Access and Usage

Developers can access the model directly on Hugging Face for integration into UI automation, accessibility testing, and GUI grounding tasks. The agentic approach enables more accurate element localization in complex, information-dense interfaces.

---

### Improve global upload performance with R2 Local Uploads
- Published: 2026-02-03T14:00:00+00:00
- Products: Cloudflare
- Tags: release, feature, performance
- URL: https://standarddb.com/cloudflare/cloudflare-launches-r2-local-uploads-in-beta-up-to-75-faster-cross-region--nqyrdzfd
- Original: https://blog.cloudflare.com/r2-local-uploads

## What's New

Cloudflare is launching **Local Uploads** for R2 in open beta. This feature automatically writes object data to a storage location close to the uploading client first, then asynchronously copies the data to the bucket's primary location. Uploaded objects are immediately accessible and remain strongly consistent throughout the replication process.

## Performance Improvements

In synthetic benchmarks and private beta testing, Local Uploads achieve **up to 75% reduction in Time to Last Byte (TTLB)** for cross-region uploads. Testing with 5 MB objects uploaded at ~20 requests per second from Western North America to an Asia-Pacific bucket showed median TTLB improvement from approximately 2 seconds to 500 milliseconds.

## How It Works

**Same Region:** If the client and bucket are in the same region, R2 follows the standard upload flow, writing directly to the bucket's storage infrastructure.

**Different Regions:** When uploading from a different region, R2 writes object data to storage near the client while publishing metadata to the bucket's primary region. The object is immediately accessible and queryable—no waiting for background replication.

## Getting Started

Enable Local Uploads via the Cloudflare Dashboard under your bucket's settings, or use Wrangler:

```shell
npx wrangler r2 bucket local-uploads enable [BUCKET]
```

## Important Limitations

Local Uploads are **not available for jurisdiction-restricted buckets** (EU, FedRAMP, or other compliance restrictions). The feature is designed for globally distributed upload workloads where performance and reliability are critical and your bucket location cannot be moved.

---

### Workflow 4.1 Beta: Event-sourced architecture
- Published: 2026-02-03T13:00:00+00:00
- Products: Vercel
- Tags: release, feature, platform
- URL: https://standarddb.com/vercel/vercel-workflow-4-1-beta-introduces-event-sourced-architecture-with-self--xvpjstkz
- Original: https://vercel.com/changelog/workflow-event-sourcing

## Event-Sourced State Management

Workflow 4.1 replaces mutable database records with an append-only event log. Instead of storing "this run is completed," the system now records the sequence of events—run_created, run_started, run_completed—and reconstructs state by replaying those events in order. Each event captures a timestamp and context, making the event log the single source of truth for workflow execution.

## Three Key Benefits

**Self-healing runs**: The new architecture automatically detects and recovers from infrastructure failures. If a queue message is lost or a race condition occurs, replaying the workflow route identifies missing state and re-enqueues necessary messages. Previously, operators had to manually intervene; now recovery is automatic.

**Complete audit trail**: Every state transition is captured in the immutable event log, enabling you to replay the exact sequence of events that led to any state. This dramatically improves debugging of distributed workflows by providing full visibility into what happened and when.

**Guaranteed consistency**: Events are append-only, so partial failures during writes cannot leave entities in an inconsistent state. The event log provides a reliable foundation that eliminates race conditions common in traditional mutable-record systems.

## Additional Improvements

Beyond the architectural shift, Workflow 4.1 adds:

- **Improved throughput**: The queue system now processes thousands of steps per second, with parallel execution when dependencies allow
- **Provider-executed tools**: `@workflow/ai` supports tools like Google Search and WebSearch that run on the model provider's infrastructure
- **NestJS support**: New `@workflow/nest` package for seamless integration with NestJS applications and dependency injection
- **Explicit Resource Management**: SWC plugin now supports TC39 resource management proposals for automatic cleanup within step and workflow functions
- **Custom serialization**: Client mode supports custom class serialization via `classes` object in manifest.json

Refer to the [Event Sourcing documentation](https://useworkflow.dev/docs/how-it-works/event-sourcing) for detailed state machine diagrams and the complete event model.

---

### Turbo build machines by default for new Pro projects
- Published: 2026-02-03T13:00:00+00:00
- Products: Vercel
- Tags: feature, platform, performance
- URL: https://standarddb.com/vercel/vercel-makes-turbo-build-machines-default-for-new-pro-projects-promises-30-70--bwdzxbzg
- Original: https://vercel.com/changelog/turbo-build-machines-by-default-for-new-pro-projects

## What's Changing

Turbo build machines are now the default configuration for all newly created Pro projects on Vercel, as well as for any Hobby projects that upgrade to the Pro plan. This change simplifies the onboarding experience and ensures teams immediately benefit from enhanced build performance without manual configuration.

## Performance Improvements

Teams that have adopted Turbo build machines report meaningful speed gains across different build scenarios:

- **Short builds (under 2 minutes):** up to 30% faster
- **Medium builds (2-10 minutes):** up to 50% faster  
- **Long builds (over 10 minutes):** up to 70% faster

These improvements stem from Turbo build machines providing 30vCPUs and 60GB of memory—a significant jump from standard build infrastructure.

## Getting Started

Existing Pro projects retain their current build machine configuration by default, but users can customize settings at any time through the [Vercel settings dashboard](https://vercel.com/d?to=%2F%5Bteam%5D%2F~%2Fsettings%2Fbuild-and-deployment%23build-machines). The feature is available across all paid plans, with documentation available for teams looking to optimize their build setup further.

---

### Zero-configuration support for Koa
- Published: 2026-02-03T13:00:00+00:00
- Products: Vercel
- Tags: platform, integration, feature
- URL: https://standarddb.com/vercel/vercel-adds-zero-configuration-support-for-koa-framework--wqjxrpng
- Original: https://vercel.com/changelog/zero-configuration-support-for-koa

## Zero-Configuration Koa Deployment

Vercel has added native support for Koa applications, the expressive HTTP middleware framework popular for building web applications and APIs. Developers can now deploy Koa apps to Vercel without any custom configuration.

## How It Works

Deploying a Koa application is straightforward. Simply write your Koa server using the framework's standard patterns and push to your repository—Vercel's platform automatically detects and deploys it:

```typescript
import Koa from 'koa'
import { Router } from '@koa/router'

const app = new Koa()
const router = new Router()

router.get('/', (ctx) => {
  ctx.body = { message: 'Hello from Koa!' }
})

app.use(router.routes())
app.use(router.allowedMethods())
app.listen(3000)
```

## Compute & Pricing

Koa applications on Vercel automatically run on **Fluid compute** with **Active CPU pricing**, meaning:

- **Automatic scaling**: Your app scales up and down based on incoming traffic
- **Pay-per-use pricing**: You only pay for the CPU time your Koa server actually uses
- **No configuration needed**: These features are enabled by default

## Getting Started

For complete deployment details and best practices, consult the [Koa on Vercel documentation](https://vercel.com/docs/frameworks/backend/koa). Deploy your first Koa app with a free Vercel account, or contact the sales team for Pro or Enterprise deployments.

---

### AI Gateway and one-click deploys now available on TRAE
- Published: 2026-02-03T13:00:00+00:00
- Products: Vercel AI SDK, Vercel
- Tags: integration, feature, platform
- URL: https://standarddb.com/vercel/vercel-s-ai-gateway-integrates-with-bytedance-s-trae-enabling-unified-model--tyvgnznr
- Original: https://vercel.com/changelog/ai-gateway-and-one-click-deploys-now-available-on-trae

## Integration Overview

ByteDance's TRAE coding agent now integrates both **Vercel AI Gateway** and direct **Vercel deployments**, bringing unified AI access and instant production shipping to over 1.6 million monthly active developers. This integration allows teams to access hundreds of models—including offerings from Anthropic, OpenAI, Google, xAI, DeepSeek, Z.AI, MiniMax, and Moonshot AI—through a single API key without managing multiple provider accounts.

## Key Features

The integration includes several developer-focused capabilities:

- **Unified model access**: Access hundreds of models across multiple providers from one API endpoint
- **Automatic failover**: Routes requests around provider outages for improved reliability
- **Zero markup pricing**: Pay provider rates directly without Vercel markup on AI tokens
- **Unified observability**: Monitor both deployments and AI usage from one dashboard
- **One-click deployment**: Deploy applications directly to Vercel production from the TRAE interface with automatic authorization

## Setup Instructions

**SOLO Mode**: Connect your Vercel account through the Integrations tab, add your AI Gateway API key, select a model, and click Deploy in the chat panel to ship to production. All projects immediately appear in your Vercel dashboard.

**IDE Mode**: Add AI Gateway as a model provider in Builder chat, select your desired model or enter a custom creator/model slug, and add your API key. Switch models with a single configuration change while maintaining unified billing through Vercel.

## Getting Started

Developers can get started with [Vercel's AI Gateway](https://vercel.com/ai-gateway) documentation and begin building with a free account. For Pro or Enterprise needs, Vercel offers personalized support and consulting.


---

### Copy visual context to agents from Vercel Toolbar
- Published: 2026-02-03T13:00:00+00:00
- Products: Vercel
- Tags: feature, integration
- URL: https://standarddb.com/vercel/vercel-toolbar-adds-copy-for-agents-to-share-visual-context-with-ai-coding--xsqstltj
- Original: https://vercel.com/changelog/copy-visual-context-to-agents

## Copy for Agents Feature

Vercel Toolbar now includes **"Copy for Agents"** functionality that enables developers to share complete visual context from their deployed applications with AI coding agents. This feature captures structured information about the current page state, layout, and component hierarchy.

## What Gets Captured

When using the copy feature, agents receive comprehensive context including:

- **Page Location**: Current URL of the deployed application
- **Viewport Information**: Screen dimensions (e.g., 1920x1080)
- **Selected Elements**: Specific text and CSS selectors for UI components
- **Component Hierarchy**: React component tree structure showing parent-child relationships
- **Original Comments**: The developer's deployment feedback or instructions

## How It Works

Instead of manually describing where changes are needed in a deployed app, developers can now copy comments directly from the Vercel Toolbar with all technical details intact. The context is structured in a format that agents can parse and act on immediately:

```
Page URL: /dashboard/projects
Viewport: 1920x1080
Selected Text: "Deploy your latest changes"
Selector: button.deploy-btn
Component Tree: App > Dashboard > ProjectList > DeployButton
```

This approach reduces communication overhead between developers and AI agents, enabling agents to understand exactly where issues occur and what changes are required without requiring additional explanation.

## Availability

The feature is available immediately to all Vercel users at no additional cost.

---

### Windsurf Tab v2: 25-75% more accepted code with Variable Aggression
- Published: 2026-02-03T12:00:00+00:00
- Products: 
- Tags: 
- URL: https://standarddb.com/windsurf/windsurf-tab-v2-25-75-more-accepted-code-with-variable-aggression--pvhdmxrs
- Original: https://windsurf.com/blog/windsurf-tab-2



---

### Introducing Deno Sandbox
- Published: 2026-02-03T12:00:00+00:00
- Products: 
- Tags: release, feature, platform, security, sdk
- URL: https://standarddb.com/deno/deno-launches-sandbox-for-running-untrusted-code-with-network-isolation-and--wbjndfby
- Original: https://deno.com/blog/introducing-deno-sandbox

## Secure Execution for LLM-Generated Code

Deno Sandbox addresses a critical new use case: platforms where large language models generate code that runs immediately without human review. This code often needs network access and API credentials, creating security risks that traditional sandboxing approaches don't fully mitigate.

The solution provides lightweight Linux microVMs running in the Deno Deploy cloud, booting in under one second. Developers can create sandboxes programmatically via JavaScript or Python SDKs, and interact with them through SSH, HTTP, or VS Code.

## Network and Secrets Protection

Two key security features prevent credential exfiltration:

- **Network Egress Control**: Restrict outbound connections to whitelisted hosts (e.g., `allowNet: ["api.openai.com"]`). Any request to unlisted hosts gets blocked at the VM boundary.
- **Secret Materialization**: API keys and credentials never appear in the environment as plaintext. Code sees only a placeholder, and the real secret materializes only when making approved outbound requests. Prompt-injected code attempting to exfiltrate the placeholder to unauthorized domains receives nothing of value.

Both features use an outbound proxy mechanism similar to coder/httpjail. Deno plans to add request inspection and modification hooks for trusted code.

## From Sandbox to Production

The `sandbox.deploy()` method deploys code directly from a sandbox to Deno Deploy production without rebuilding or re-authenticating:

```
const build = await sandbox.deploy("my-app", {
  production: true,
  build: { mode: "none", entrypoint: "server.ts" },
});
```

## Storage and Persistence

- **Volumes**: Read-write storage for databases, caches, and user data
- **Snapshots**: Read-only base images for pre-installed toolchains; create fresh development environments in seconds

## Pricing and Availability

Deno Sandbox launches in beta with usage-based pricing integrated into Deno Deploy plans:

- $0.05/h CPU time (40 hours included with Pro)
- $0.016/GB-h memory (1000 GB-h included with Pro)
- $0.20/GiB-month volume storage (5 GiB included with Pro)

Sandboxes run with 2 vCPUs, 768 MB–4 GB memory, and a maximum 30-minute lifetime. Use cases include AI agents, secure plugin systems, ephemeral CI runners, and customer-supplied code execution.

---

### Deno Deploy is Generally Available
- Published: 2026-02-03T12:00:00+00:00
- Products: 
- Tags: release, platform, feature, api
- URL: https://standarddb.com/deno/deno-deploy-reaches-general-availability-with-zero-config-deployment-and--kgfkbglh
- Original: https://deno.com/blog/deno-deploy-is-ga

## Deno Deploy Goes Generally Available

Deno Deploy, a platform for deploying JavaScript and TypeScript applications, has reached general availability. The service eliminates traditional deployment friction by supporting any JavaScript framework—including Next.js, Astro, and SvelteKit—with automatic framework detection and build command execution.

## Key Features and Developer Experience

**Zero-Config Continuous Deployment**: Connect a GitHub repository to automatically get continuous deployment with live previews for every commit, isolated databases for pull requests, and easy promotion to production. A new `deno deploy` CLI subcommand provides manual deployment control for CI/CD pipelines and local testing.

**Integrated Databases**: The platform now supports both Deno KV (global key-value store) and Postgres databases. Through a partnership with Prisma, users can provision new databases directly from the dashboard—including free databases for each pull request. Environment variables are automatically managed, ensuring the same application code works across all contexts.

**Local Development with `--tunnel`**: A new flag bridges the local-to-production gap by automatically pulling centralized environment variables, running a local dev server with a public shareable URL, and sending telemetry to Deno Deploy.

## Automatic Observability and Deno Sandbox

All projects hosted on Deno Deploy automatically capture logs, traces, and metrics—including `console.log`, fetch requests, HTTP events, V8 events, garbage collection, and I/O. Logs are automatically associated with requests for easier debugging.

**Deno Sandbox**, a new service within Deno Deploy, provides isolated microVMs that boot in under a second for programmatically executing code securely. This is designed to support AI-generated code execution with full isolation and prompt-injection resistant secrets handling.

## Pricing

The free tier includes one million requests per month, 100 GB of egress, and 15 CPU hours. Pro plans offer flexible scaling, with custom enterprise pricing available for teams with advanced security and performance requirements.

---

### Supabase is now an official Claude connector
- Published: 2026-02-03T07:00:00+00:00
- Products: Supabase
- Tags: integration, feature, api
- URL: https://standarddb.com/supabase/supabase-launches-official-claude-connector-with-32-database-management-tools--jxkmncrg
- Original: https://supabase.com/blog/supabase-is-now-an-official-claude-connector

## Official Claude Integration

Supabase has been added to Claude's official connectors, enabling seamless integration between the database platform and Anthropic's AI assistant. Developers can now authorize Claude to access their Supabase organization and manage their entire infrastructure through conversational prompts.

## Available Capabilities

The Supabase MCP server includes **32 tools** spanning multiple platform features:

- **Database Management**: Execute SQL queries on PostgreSQL databases, design and modify table schemas, list tables and extensions, generate TypeScript types for your schema, and receive security recommendations
- **Project Operations**: List and create projects, get cost estimates, pause and restore projects, and access real-time logs
- **Branching and Migrations**: Create development branches, merge changes, rebase and reset branches, and manage migrations
- **Storage and Edge Functions**: Manage storage buckets, deploy serverless functions, and handle related configurations
- **Documentation**: Search Supabase documentation directly from Claude

## Getting Started

To enable the connector, open Claude.ai or Claude Desktop, access the connectors menu, and select Supabase to authorize Claude access to your organization. The integration works on all Supabase plans and is available to users of paid Claude plans (Pro, Team, Enterprise, Max).

Common use cases include asking Claude to check security advisors and fix issues, request schema changes for automatic execution, and deploy edge functions with a single prompt—all without leaving the conversation interface.

---

### GitHub Mobile: Comment on unchanged lines in pull request files
- Published: 2026-02-03T00:00:00+00:00
- Products: GitHub
- Tags: feature, platform, release
- URL: https://standarddb.com/github/github-mobile-now-supports-commenting-on-unchanged-lines-in-pull-request-diffs--vmzrrqwc
- Original: https://github.blog/changelog/2026-02-03-github-mobile-comment-on-unchanged-lines-in-pull-request-files

## Overview

GitHub Mobile has enhanced its pull request review capabilities with the ability to comment on unchanged lines within diff views. Previously, the mobile app limited comments to lines directly surrounding actual code changes, restricting reviewers' ability to provide comprehensive feedback.

## What's New

- **Expanded commenting scope**: Reviewers can now expand diffs to reveal unchanged code and leave comments or suggested changes anywhere in the file
- **Bidirectional comment display**: Comments added on unchanged lines via github.com are also visible and editable in GitHub Mobile
- **Full context preservation**: Reviewers maintain access to surrounding code context without losing visibility of the actual changes

## Benefits for Developers

This feature enables more thorough code reviews by allowing:
- Flagging changes that should have been made but weren't
- Suggesting updates and improvements beyond the immediate diff context
- Adding helpful context for reviewers on unchanged parts of a file
- Maintaining consistency between web and mobile review experiences

## Availability

The update is immediately available on the latest production builds of GitHub Mobile for both [iOS](https://apps.apple.com/us/app/github/id1477376905) and [Android](https://play.google.com/store/apps/details?id=com.github.android).

---

### R2 - Improve Global Upload Performance with R2 Local Uploads - Now in Open Beta
- Published: 2026-02-03T00:00:00+00:00
- Products: Cloudflare
- Tags: feature, performance, release
- URL: https://standarddb.com/cloudflare/cloudflare-r2-local-uploads-enters-open-beta-reduces-upload-latency-by-up-to-75--mczbfdlf
- Original: https://developers.cloudflare.com/changelog/2026-02-03-r2-local-uploads

## R2 Local Uploads Now Available in Open Beta

Cloudflare has introduced **Local Uploads** for R2 object storage, a new feature designed to optimize upload performance for users with globally distributed traffic. The feature is now available in open beta.

### How It Works

With Local Uploads enabled, object data is written to storage infrastructure geographically near the uploading client, then asynchronously replicated to your primary R2 bucket. This approach eliminates the performance penalty of uploading from distant regions while maintaining **strong consistency** throughout the process—objects are immediately accessible after the initial write.

### Performance Gains

In Cloudflare's testing, Local Uploads delivered **up to 75% reduction in Time to Last Byte (TTLB)** for upload requests. The benchmark showed median latency dropping from approximately 2 seconds to 500ms, a significant improvement for applications where upload speed is critical.

### Ideal Use Cases

Local Uploads is particularly valuable when:
- Your user base spans multiple geographic regions
- Upload performance and reliability are critical to your application
- You want to optimize write performance without relocating your bucket

### Getting Started

Enabling Local Uploads is seamless with no downtime or traffic interruption. You can enable it through the Cloudflare Dashboard or via the Wrangler CLI:

```
npx wrangler r2 bucket local-uploads enable <BUCKET_NAME>
```

### Pricing

There is **no additional cost** to enable Local Uploads. Upload requests incur standard Class A operation costs, the same as regular uploads. Existing uploads will continue to complete normally during the transition.

---

### Apple's Xcode now supports the Claude Agent SDK
- Published: 2026-02-03T00:00:00+00:00
- Products: Claude Code, Claude
- Tags: release, feature, sdk, integration, api
- URL: https://standarddb.com/anthropic/xcode-26-3-integrates-claude-agent-sdk-for-autonomous-coding-tasks-within-apple--zfbqyjfw
- Original: https://www.anthropic.com/news/apple-xcode-claude-agent-sdk

## Claude Agent SDK Now Native to Xcode

Xcode 26.3 introduces a direct integration with the Claude Agent SDK, bringing the full capabilities of Claude Code to Apple's development environment. This upgrade moves beyond the turn-by-turn assistance available in previous versions, enabling autonomous work on sophisticated coding tasks for iPhone, iPad, Mac, Apple Watch, Apple Vision Pro, and Apple TV platforms.

## Key Capabilities

The integration delivers several powerful features for Apple platform developers:

- **Visual Verification with Previews**: Claude can capture and analyze Xcode Previews to verify visual output, identify issues with SwiftUI implementations, and iterate autonomously. This creates a feedback loop where Claude can validate designs against developer intent and refine interfaces.
  
- **Project-Wide Reasoning**: Rather than working in isolation on a single file, Claude can explore full project structure and understand how frameworks like SwiftUI, UIKit, and Swift Data interconnect. This architectural awareness enables more informed code changes across the entire app.
  
- **Autonomous Task Execution**: Developers can specify high-level goals rather than detailed instructions. Claude breaks down tasks independently, determines which files to modify, executes changes, and iterates if needed. The integration includes access to Apple's documentation for framework-specific guidance.
  
- **Model Context Protocol Support**: Beyond direct IDE integration, Xcode 26.3 exposes capabilities through the Model Context Protocol, allowing Claude Code users to tap into Xcode's features from the command line.

## Availability

Xcode 26.3 is available immediately as a release candidate for all Apple Developer Program members, with a public App Store release coming soon.

---

### Agents, Workflows - Agents SDK v0.3.7: Workflows integration, synchronous state, and scheduleEvery()
- Published: 2026-02-03T00:00:00+00:00
- Products: Cloudflare Workers
- Tags: release, feature, api, sdk, platform
- URL: https://standarddb.com/cloudflare/cloudflare-agents-sdk-v0-3-7-adds-workflows-integration-and-synchronous-state--xvzybsyq
- Original: https://developers.cloudflare.com/changelog/2026-02-03-agents-workflows-integration

## Cloudflare Workflows Integration

Agents SDK v0.3.7 now provides first-class support for **Cloudflare Workflows**, creating a powerful combination where agents handle WebSocket connections and real-time state management while workflows manage durable execution, retries, and human-in-the-loop approval flows.

The new `AgentWorkflow` class allows you to define typed workflows with direct access to your agent:

- Define workflows that can call agent methods via RPC and merge agent state
- Use the `step` API for durable, idempotent operations that won't repeat on retry
- Call non-durable methods like `broadcastToClients()` for real-time progress reporting

Start workflows directly from your agent using `runWorkflow()` and handle lifecycle events with callbacks like `onWorkflowProgress()`, `onWorkflowComplete()`, and `onWorkflowError()`. Additional workflow control methods include `approveWorkflow()`, `pauseWorkflow()`, and `terminateWorkflow()` for human-in-the-loop and orchestration patterns.

## Synchronous State Management and Task Scheduling

State updates are now **synchronous** with a new `validateStateChange()` hook that lets you reject or transform state changes before they're applied. This provides immediate feedback and validation without asynchronous round trips.

The SDK also introduces `scheduleEvery()` for fixed-interval recurring tasks with built-in overlap prevention, making it simple to schedule operations like data syncs without complex cron configuration.

## Callable System Improvements

The callable system now supports:

- **Client-side RPC timeout** — Specify timeout values on callable method invocations
- **Stream error signaling** — Use `StreamingResponse.error()` for graceful error handling in streaming responses
- **Method introspection** — Discover available callable methods with `getCallableMethods()`
- **Connection close handling** — Pending calls are automatically rejected when clients disconnect

## Security and Routing Enhancements

Email replies are now secured with **HMAC-SHA256 signed headers** to prevent unauthorized routing of emails to agent instances. A new `basePath` option allows custom URL routing by bypassing default URL construction.

---

### The Dependabot Proxy is now open source with an MIT license
- Published: 2026-02-03T00:00:00+00:00
- Products: GitHub Dependabot, GitHub
- Tags: open-source, release, feature, security, integration
- URL: https://standarddb.com/github/github-releases-dependabot-proxy-as-open-source-under-mit-license--zjhqmwnz
- Original: https://github.blog/changelog/2026-02-03-the-dependabot-proxy-is-now-open-source-with-an-mit-license

## Now Open Source

The [Dependabot Proxy](https://github.com/dependabot/proxy), previously internal to GitHub, is now available as open source under the MIT license. This HTTP proxy is a critical component of the Dependabot dependency management system, handling authentication when Dependabot connects to the GitHub API and private package registries.

## What Developers Can Do

With the source code now public, developers can:

- **Review authentication logic** to understand how Dependabot securely handles credentials across different package managers and Git servers
- **Submit bug fixes and ecosystem support** to expand the proxy's capabilities
- **File issues and collaborate** directly with GitHub's development team in an open repository

The proxy supports nine major package ecosystems: **npm, Maven, Docker, Cargo, Helm, NuGet, pip, RubyGems, and Terraform**, along with Git servers including GitHub, Azure DevOps, and others.

## Compliance and Transparency

This move is particularly valuable for organizations with strict compliance and supply chain security requirements. Teams can now audit the exact mechanisms used in their software supply chain and verify that authentication follows their security policies. This aligns with broader industry trends toward transparency in dependency management tools.

The proxy was built in Go and is part of a larger open source ecosystem that includes [dependabot-core](https://github.com/dependabot/dependabot-core), the engine powering all Dependabot features.



---

### Dependabot now supports OIDC authentication
- Published: 2026-02-03T00:00:00+00:00
- Products: GitHub Dependabot
- Tags: feature, security, api
- URL: https://standarddb.com/github/dependabot-adds-oidc-authentication-support-for-private-registries--xvpnsbkl
- Original: https://github.blog/changelog/2026-02-03-dependabot-now-supports-oidc-authentication

## OIDC Authentication Now Available

Dependabot can now use OpenID Connect (OIDC) to authenticate with private registries, eliminating the security risks associated with storing long-lived credentials as repository secrets. This authentication method mirrors GitHub Actions' existing OIDC federation approach, allowing Dependabot update jobs to dynamically obtain short-lived credentials from your cloud identity provider.

## Supported Registries

The following private registries now support OIDC-based authentication:

- **AWS CodeArtifact**
- **Azure DevOps Artifacts**
- **JFrog Artifactory**

## Key Benefits

**Enhanced Security:** Removes static, long-lived credentials from repositories. Short-lived, dynamically generated tokens significantly reduce the operational overhead and attack surface.

**Simplified Management:** Enables secure, policy-compliant access to private registries without manual credential rotation.

**Rate Limiting Relief:** Dynamic credentials help teams avoid hitting rate limits that are often associated with static tokens.

## Getting Started

To enable OIDC authentication, update your `dependabot.yml` configuration to use the new OIDC authentication type for your supported registry. Refer to the GitHub documentation on [configuring access to private registries for Dependabot](https://docs.github.com/code-security/dependabot/working-with-dependabot/configuring-access-to-private-registries-for-dependabot) for detailed setup instructions and configuration examples.

---

### Simplified Copilot model enablement experience for individual users
- Published: 2026-02-03T00:00:00+00:00
- Products: GitHub Copilot
- Tags: feature, platform, release
- URL: https://standarddb.com/github/github-copilot-simplifies-model-selection-with-automatic-discovery-for-pro-and--wtctzbnl
- Original: https://github.blog/changelog/2026-02-03-simplified-copilot-model-enablement-experience-for-individual-users

## Simpler Model Access, No Settings Required

GitHub has simplified the Copilot experience by eliminating friction around discovering and enabling new models. Previously, users had to navigate settings pages or manage policies to access newly released models. Now, as GitHub rolls out new models, they become immediately available within Copilot's interface across all subscription tiers.

## Key Changes

- **Instant availability**: New models appear directly in Copilot without extra steps, clicks, or configuration
- **Auto mode option**: Users can let GitHub select the best model automatically or manually choose one themselves
- **Transparent data policies**: All existing data-usage policies remain unchanged and continue to apply
- **User control preserved**: Users still maintain full control over which models they use and when

## What Developers Need to Know

This is a user experience improvement rather than a breaking change. Copilot Pro+, Pro, and Free users will see new models available for selection within their normal Copilot workflow. If you're using Copilot, check back over the next day as the rollout completes. No action is required—the change is opt-in, and your existing preferences and policies remain intact.

---

### February 3, 2026
- Published: 2026-02-03T00:00:00+00:00
- Products: 
- Tags: feature, model, api
- URL: https://standarddb.com/deepgram/deepgram-adds-hebrew-farsi-and-urdu-support-to-nova-3-speech-to-text-model--yybschtz
- Original: https://developers.deepgram.com/changelog/2026/2/3

## Nova-3 Language Expansion

Deepgram has expanded its Nova-3 speech-to-text model with support for three new languages, enabling developers to reach a broader global audience.

### Newly Supported Languages

The Nova-3 model now includes monolingual support for:
- **Hebrew** (language code: `he`)
- **Persian/Farsi** (language code: `fa`)
- **Urdu** (language code: `ur`)

### What This Means for Developers

Developers can now integrate Deepgram's industry-leading speech-to-text capabilities into applications targeting Hebrew, Farsi, and Urdu-speaking communities. These additions make it easier to build more inclusive voice experiences without needing separate third-party solutions.

### Getting Started

The new language models are available immediately through the standard Deepgram API. Developers can specify the desired language using the appropriate language codes when making requests. Documentation is available on the [Models and Language Overview](https://developers.deepgram.com/docs/models-languages-overview) page.

---

### Security Center - Threat actor identification with "also known as" aliases
- Published: 2026-02-03T00:00:00+00:00
- Products: Cloudflare
- Tags: feature, api, security
- URL: https://standarddb.com/cloudflare/cloudflare-security-center-adds-threat-actor-alias-lookup-to-simplify-cross--wxzjxbqf
- Original: https://developers.cloudflare.com/changelog/2026-02-03-threat-actor-name-mapping

## What's New

Cloudflare has added an **"Also known as"** field to Threat Events in Security Center, making it easier to identify threat actors across different security vendor naming conventions.

## Key Features

- **Dashboard Integration**: View threat actor aliases directly in the Threat Events dashboard by expanding event details under the **Attacker** field
- **Customizable Table View**: Add the new field as a column in your configurable table view for quick reference
- **API Access**: The alias data is available via the Threat Events API for programmatic access

## Benefits for Security Teams

This addition addresses a common challenge in threat intelligence research: different security vendors often use different names for the same threat groups, making correlation difficult. With Cloudflare's standardized alias field, you can:

- **Quickly map** Cloudflare-tracked actors to naming conventions used by other vendors without manual research
- **Streamline cross-referencing** by immediately identifying if a detected threat actor matches groups already being monitored in other intelligence feeds
- **Reduce operational overhead** by consolidating threat actor research across multiple sources

## Getting Started

For implementation details and API integration, refer to the [Threat Events API documentation](https://developers.cloudflare.com/api/resources/cloudforce_one/subresources/threat_events/).

---

### Python 3.13 and 3.14 are now available 
- Published: 2026-02-02T13:00:00+00:00
- Products: Vercel
- Tags: release, platform
- URL: https://standarddb.com/vercel/vercel-adds-python-3-13-and-3-14-support-3-14-becomes-default-in-coming-months--qzcxxdxr
- Original: https://vercel.com/changelog/python-3-13-and-3-14-are-now-available

## Python 3.13 and 3.14 Now Available on Vercel

Vercel has expanded its Python runtime support for **Builds and Functions**, now offering Python 3.13 and Python 3.14 alongside the previously supported Python 3.12. This allows developers to leverage the latest Python features and improvements for their serverless workloads.

### Default Runtime Transition

Projects without an explicitly specified Python version will continue using Python 3.12 as the default for now. However, **the default will switch to Python 3.14 in the coming months**. Developers who need to maintain Python 3.12 should proactively specify their required version in their project manifest to avoid unexpected runtime changes.

### How to Pin Your Python Version

To lock your project to Python 3.12, specify the version constraint in your project configuration:

- **For `pyproject.toml`**: Add `requires-python = "~=3.12.0"` under the `[project]` section
- **For `Pipfile`**: Add `python_version = "3.12"` under the `[requires]` section

Refer to [Vercel's Python documentation](https://vercel.com/docs/functions/runtimes/python) for additional configuration details and best practices.

---

### Email security - Improved Accessibility and Search for Monitoring
- Published: 2026-02-02T11:05:33+00:00
- Products: Cloudflare
- Tags: feature, platform
- URL: https://standarddb.com/cloudflare/cloudflare-email-security-improves-monitoring-dashboard-with-search--fdvzlbxr
- Original: https://developers.cloudflare.com/changelog/2026-02-02-improved-accessibility-search-for-monitoring

## Monitoring Dashboard Updates

Cloudflare has enhanced its Email Security monitoring experience with a comprehensive set of UI and accessibility improvements designed to help administrators gain clearer insights into email security events.

## Key Improvements

**Visual and Data Enhancements:**
- Optimized contrast and introduced stacked bar charts for clearer data visualization and trend analysis
- New granular data breakdowns including dispositions by month, malicious email details, link actions, and impersonation tracking

**Accessibility & Usability:**
- **Widget Search**: Added search functionality across multiple widgets including Policies, Submitters, and Impersonation, making it faster to find relevant data
- **Actionable UI**: All available actions are now accessible via dedicated buttons for better discoverability
- **State Indicators**: Improved UI states that clearly communicate loading, empty datasets, and error conditions

## Availability

These improvements apply to all Email Security packages:
- Advantage
- Enterprise
- Enterprise + PhishGuard

The update rolls out immediately to all qualifying customers with no action required.

---

### Snowflake and OpenAI partner to bring frontier intelligence to enterprise data
- Published: 2026-02-02T06:00:00+00:00
- Products: OpenAI
- Tags: 
- URL: https://standarddb.com/openai/snowflake-and-openai-partner-to-bring-frontier-intelligence-to-enterprise-data--svvsznwd
- Original: https://openai.com/index/snowflake-partnership



---

### Using Agents to Scale Harvey's Knowledge Sources
- Published: 2026-02-02T00:00:00+00:00
- Products: 
- Tags: feature, platform, api
- URL: https://standarddb.com/harvey/harvey-scales-legal-knowledge-coverage-to-60-jurisdictions-with-autonomous--fwsfjrnn
- Original: https://www.harvey.ai/blog/using-agents-to-scale-harveys-knowledge-sources

## Automated Discovery and Integration at Scale

Harvey has introduced **The Data Factory**, an autonomous pipeline that dramatically reduces the manual work required to add new legal knowledge sources to its platform. Previously, integrating a new jurisdiction like Brazilian federal case law required weeks of manual effort: identifying repositories, building custom connectors, ingesting data, hand-labeling test cases, and recruiting domain experts for quality review. Now, this process is largely automated.

## Architecture and Core Components

The system operates through three main stages:

- **Intake Engine**: Automatically discovers legal sources through jurisdiction mapping and compliance review, converting customer requests and coverage gaps into vetted, pipeline-ready data sources
- **Evaluation Pipeline**: Tests whether agents can effectively use new sources to solve real legal problems using synthetic scenario generation, production simulation, and multi-agent quality assessment
- **Configuration Layer**: Defines jurisdictions declaratively with domain lists, filter hierarchies, permissions, and agent-specific instructions

## Impact and Scale

Since its launch in August 2025, The Data Factory has expanded Harvey's knowledge coverage from 6 to 60+ jurisdictions and integrated 400+ unique legal data sources. This automation enables the platform to respond rapidly to customer needs—a São Paulo customer requesting analysis of a Brazilian federal ruling that would previously have required weeks of preparation was served successfully within 72 hours through sources the automated pipeline had already indexed and validated.

---

### February 2, 2026
- Published: 2026-02-02T00:00:00+00:00
- Products: 
- Tags: release, feature, api, sdk, model
- URL: https://standarddb.com/elevenlabs/elevenlabs-ships-eleven-v3-ga-wav-support-and-agents-platform-enhancements--yytnnhrr
- Original: https://elevenlabs.io/docs/changelog/2026/2/2

## Eleven v3 General Availability

ElevenLabs' Eleven v3 model has exited alpha and is now generally available with improvements to stability, accuracy, and latency compared to previous versions.

## Text-to-Dialogue Updates

The Text-to-Dialogue API now supports WAV output formats alongside existing MP3, PCM, and OPUS options:
- WAV formats available: 8000, 16000, 22050, 24000, 32000, 44100, and 48000 Hz sample rates
- **Pro tier requirement**: WAV formats with 44.1kHz sample rate require a Pro subscription or above

## Agents Platform Enhancements

Multiple new capabilities have been added for agent developers:

- **Branch management**: Agent branches can now be renamed via the Update branch endpoint using the new `name` field (1-140 characters)
- **Safety controls**: Added `AlignmentGuardrail` type to `GuardrailsV1` schema for conversation safety guardrails
- **Turn-taking control**: New `speculative_turn` field allows fine-tuning of agent turn-taking behavior
- **Procedure references**: Agent patches now support `procedure_refs` for referencing reusable procedures
- **Webhook filtering**: Response filtering capabilities via `response_filter_mode` and `response_filters` fields
- **Debugging**: Added `raw_error_message` field to webhook tool events for improved error visibility
- **Testing**: New `check_any_tool_matches` option allows relaxed tool call matching during agent testing
- **Secrets management**: Get workspace secrets endpoint now supports pagination with `page_size` (max 100) and `cursor` parameters

## Workspace and Permissions Updates

Workspace endpoints now clarify specific permission requirements: `group_members_manage` for group operations and `WORKSPACE_MEMBERS_INVITE` for workspace invitations. New permission types have been added to support these granular controls.

## SDK Releases

All major SDKs have been updated with the new features:
- **Python SDK v2.34.0**: WAV formats, webhook filtering, branch renaming, pagination, speculative turns
- **JavaScript SDK v2.34.0**: Same feature parity as Python
- **React/Client SDKs v0.14.0-beta.0**: Reduced audio chunk latency from 250ms to 100ms
- **Widget packages**: Multiple releases addressing microphone permission handling, UI sync issues, and Tailwind v4 updates

Additional bugfixes address URL streaming in Scribe, agent initialization, and microphone permission recovery across all SDK variants.

---

### WAF - WAF Release - 2026-02-02
- Published: 2026-02-02T00:00:00+00:00
- Products: Cloudflare
- Tags: release, security, feature
- URL: https://standarddb.com/cloudflare/cloudflare-waf-adds-detections-for-django-sql-injection-and-xwiki-rce--cctczhdh
- Original: https://developers.cloudflare.com/changelog/2026-02-02-waf-release

## New WAF Detections Released

Cloudflare's Web Application Firewall (WAF) now includes protection against two significant vulnerabilities discovered in February 2026.

### CVE-2025-64459: Django SQL Injection

Django versions prior to 5.1.14, 5.2.8, and 4.2.26 contain a SQL injection vulnerability. The flaw exists in QuerySet methods and the `Q()` class when processing crafted dictionaries, allowing attackers to execute arbitrary SQL queries. The new detection rule automatically blocks requests attempting to exploit this vulnerability.

### CVE-2025-24893: XWiki Remote Code Execution

XWiki installations are vulnerable to unauthenticated remote code execution through the SolrSearch endpoint. This critical vulnerability affects entire XWiki installations regardless of authentication status. The WAF now blocks malicious requests targeting this endpoint by default.

### Rule Updates

| Rule | Detection | New Action |
|------|-----------|-----------|
| XWiki RCE (CVE-2025-24893) | Remote Code Execution | Block |
| Django SQL Injection (CVE-2025-64459) | SQL Injection via QuerySet | Block |

Additionally, Cloudflare refined the metadata description for the NoSQL/MongoDB SQL injection detection rule while keeping its blocking behavior unchanged.

**Action Items:** Review your WAF rules to ensure these protections are enabled. If you're running affected Django or XWiki versions, prioritize patching while the WAF rules provide defense-in-depth protection.

---

### Introducing the Codex app
- Published: 2026-02-02T00:00:00+00:00
- Products: OpenAI API, ChatGPT
- Tags: release, feature, platform, api, integration, open-source
- URL: https://standarddb.com/openai/openai-launches-codex-app-for-macos-with-multi-agent-orchestration-and-skill--rsbvfwvv
- Original: https://openai.com/index/introducing-the-codex-app

## Codex App for macOS Now Available

OpenAI has introduced the Codex app, a dedicated desktop application designed to serve as a command center for managing multiple coding agents simultaneously. The app addresses a fundamental shift in how developers work with AI—moving from single-agent code generation to orchestrating coordinated teams of agents across the full software development lifecycle, from design through maintenance.

## Multi-Agent Orchestration and Parallel Execution

The Codex app enables developers to:

- **Run agents in parallel** across separate threads organized by project, allowing seamless context switching without losing conversation history
- **Manage code changes collaboratively** by reviewing agent modifications in-thread, commenting on diffs, and opening changes in your editor for manual refinement
- **Avoid merge conflicts** with built-in worktree support, allowing multiple agents to work on the same repository simultaneously on isolated code copies
- **Integrate existing workflows** by picking up session history and configuration from the Codex CLI and IDE extensions

## Skills Framework Extends Agent Capabilities

Beyond code generation, the new skills framework allows Codex to tackle complex tasks requiring information synthesis, problem-solving, writing, and workflow automation. Skills bundle instructions, resources, and scripts that enable Codex to reliably connect to external tools and execute multi-step processes. The Codex app includes a dedicated interface to create and manage skills, with developers able to explicitly request specific skills or let the agent choose automatically.

OpenAI demonstrated this capability by having Codex build a complete 3D racing game using over 7 million tokens, combining image generation and game development skills to handle design, development, and QA testing independently.

## Expanded Access and Rate Limit Increases

For a limited time, Codex is included with **ChatGPT Free and Go** plans, significantly expanding access to the technology. Additionally, OpenAI is **doubling rate limits** across ChatGPT Plus, Pro, Business, Enterprise, and Edu plans—improvements that apply across the desktop app, CLI, IDE extensions, and cloud deployments.

## Open Source Skills Library

OpenAI is releasing an open source library of skills built internally for common developer workflows, including:

- **Figma integration** for translating designs into production-ready UI code with visual parity
- **Project management** capabilities for bug triage and release tracking
- **Evaluation and training workflows** for ML systems

Developers can fork and customize these skills or contribute new ones to the community repository.

---

### Introducing the Codex app
- Published: 2026-02-02T00:00:00+00:00
- Products: OpenAI
- Tags: release, feature, platform, api
- URL: https://standarddb.com/openai/openai-launches-codex-app-a-multi-agent-development-environment-with-native--wctsfzwz
- Original: https://www.youtube.com/watch?v=HFM3se4lNiw

## Overview

OpenAI has introduced the Codex app, a native macOS application designed to streamline AI-assisted software development. The app serves as a centralized command center where developers can create projects, assign tasks to AI agents, supervise their work in real-time, and manage multiple parallel development streams.

## Key Features

**Project & Task Management**
- View all projects in a unified sidebar with live task tracking
- Monitor multiple agents working in parallel without conflicts
- Review completed and in-progress work from a single interface

**Code Review & Iteration**
- View diffs of agent-generated changes directly in the app
- Leave inline comments for feedback and request iterations
- Merge approved changes or open modifications in Xcode for deeper inspection
- Build and run applications directly from the Codex interface

**Skills System**
- Package custom tools and conventions into reusable "skills" that agents can leverage
- Pre-built integrations (e.g., Figma skill for design-to-code workflows using MCP)
- Create team-specific skills to fit existing workflows

**Automations & Background Work**
- Schedule long-running tasks to execute in the background at specified intervals
- Examples include triaging Sentry alerts or Linear bugs automatically
- Option to delegate resource-intensive work to cloud infrastructure with the same interface

**Isolated Development Environments**
- Worktrees allow each agent to work on isolated code copies, preventing conflicts
- Enables safe experimentation without breaking production setups

## Availability & Roadmap

The Codex app is available for download on macOS immediately. Windows support is coming soon. Developers can access it at openai.com/codex.

## Significance

This represents a shift in how developers interact with AI coding tools, moving from chat-based interfaces to a proper development environment with supervision, review, and automation capabilities. The app targets the full development workflow rather than isolated code generation tasks.

---

### v0.5.79
- Published: 2026-02-01T06:06:55+00:00
- Products: Sim Studio
- Tags: release, feature, performance, bugfix, platform
- URL: https://standarddb.com/sim/sim-v0-5-79-adds-enrichment-tools-increases-rate-limits-across-plans--hdyjpycz
- Original: https://github.com/simstudioai/sim/releases/tag/v0.5.79

## Features & Improvements

**New Tools**: Added enrichment functionality to expand the toolset available to agents and workflows.

**Rate Limit Increases**: Increased rate limits across all subscription plans alongside improved socket connections. Users will now receive reconnection notifications when socket connections are interrupted, improving visibility into connection state changes.

## Bug Fixes & Optimizations

**Workflow Optimization**: Improved loop and parallel regeneration performance while preventing duplicate agent tools from being created during workflow execution.

**MCP Tool Timeout Handling**: Fixed an issue where the Model Context Protocol (MCP) SDK's `callTool` method now properly respects custom timeout configurations, overriding the previous 60-second default timeout.

## What Developers Need to Know

If you're building agents or workflows with Sim Studio, upgrade to take advantage of the new enrichment tools. The rate limit increases mean you can make more API calls per minute depending on your plan tier. If you're using MCP integrations, you can now configure tool timeout durations without being constrained by the 60-second default.

---

### v0.5.78
- Published: 2026-01-31T21:58:16+00:00
- Products: Sim Studio
- Tags: release, feature, bugfix, integration
- URL: https://standarddb.com/sim/sim-launches-v0-5-78-with-google-maps-and-dspy-integration-plus-bug-fixes--pjhvstfc
- Original: https://github.com/simstudioai/sim/releases/tag/v0.5.78

## New Integrations

Sim v0.5.78 introduces two significant new tools for developers building AI workflows:

- **Google Maps integration**: Access location-based data and mapping capabilities within your Sim workflows
- **DSPy support**: Integrate the DSPy framework for building and optimizing language model pipelines

## Developer Experience Improvements

The release migrates all tool files to use absolute imports, improving code consistency and maintainability across the SDK. Additionally, a new invitations query hook has been added to simplify permission and access management workflows.

## Bug Fixes and Stability

This patch addresses several important issues:

- **Visibility controls**: Updated parameter visibility settings to allow both users and LLMs to access non-sensitive tool parameters, expanding what AI agents can do
- **Executor improvements**: Fixed conditional logic handling within parallel execution contexts
- **MCP timeout**: Increased timeout from 1 minute to 10 minutes for more reliable long-running operations
- **Billing**: Corrected plan detection from Stripe subscription objects
- **Editor permissions**: Fixed advanced toggle to properly respect user edit permissions

## Getting Started

Update to v0.5.78 to access the new Google Maps and DSPy tools, and benefit from improved stability and permission handling.

---

### v0.5.77
- Published: 2026-01-30T23:18:40+00:00
- Products: Sim Studio
- Tags: release, feature, integration, bugfix, performance
- URL: https://standarddb.com/sim/sim-studio-releases-v0-5-77-with-similarweb-and-cal-com-integrations--xybyygfx
- Original: https://github.com/simstudioai/sim/releases/tag/v0.5.77

## New Integrations

Sim Studio v0.5.77 introduces two new tool integrations to expand workflow capabilities:

- **SimilarWeb**: Added as a new tool integration for web analytics and competitive intelligence workflows
- **Cal.com**: Integrated for calendar and scheduling functionality within automation workflows

## UX Improvements

Several user experience enhancements were included in this release:

- **Note Block Dragging**: Enabled body dragging on note blocks to match the behavior of workflow blocks, improving consistency across the interface
- **Deployment Descriptions**: Human-readable version descriptions for deployments for clearer tracking and documentation
- **Documentation**: Added instant copy button to docs and performance optimizations for faster page loads

## Bug Fixes & Infrastructure

The release addresses critical stability and compatibility issues:

- **Streaming**: Fixed multi-byte UTF-8 character handling when split across chunks
- **Token Management**: Corrected Anthropic token limits for streaming with tool calls
- **Agent Logging**: Fixed agent cost filtering in trace spans
- **Executor Logic**: Improved conditional deactivation for loops and parallel operations
- **Redis Migration**: Migrated sockets to Redis for improved scalability

## Docker & Validation Updates

Docker Compose environment variables were updated, and hosted API key and credential validation for the Copilot feature were enhanced.

---

### Vercel Sandboxes are now generally available
- Published: 2026-01-30T13:00:00+00:00
- Products: Vercel
- Tags: release, feature, platform, sdk
- URL: https://standarddb.com/vercel/vercel-sandboxes-now-generally-available-for-safely-executing-untrusted-code--vsxdrxnb
- Original: https://vercel.com/changelog/vercel-sandboxes-ga

## What's New

Vercel Sandboxes are now generally available as a production-ready service for executing untrusted code in isolated, ephemeral environments. Each sandbox runs inside a Firecracker microVM, providing strong security isolation without exposing production systems to risk.

## Key Features

- **Isolated Execution**: Code running in sandboxes is blocked from accessing environment variables, database connections, and cloud resources
- **Ephemeral Compute**: Sandboxes are designed to be short-lived, spinning up and down as needed
- **Use Cases**: Ideal for running AI agent-generated outputs, unverified user uploads, and third-party code

## Getting Started

Vercel provides both SDK and CLI options for developers:

**SDK Usage** (Node.js):
```typescript
import { Sandbox } from '@vercel/sandbox';

const sandbox = await Sandbox.create();
await sandbox.runCommand({
  cmd: 'node',
  args: ["-e", 'console.log("Hello from Vercel Sandbox!")'],
  stdout: process.stdout,
});
await sandbox.stop();
```

**CLI**: Launch an interactive shell with `npx sandbox create --connect`

## Production Adoption

Sandboxes are already in production use by notable teams including v0, Blackbox AI, and RooCode, validating the reliability and utility of the offering.

Developers can explore the full [documentation](https://vercel.com/docs/vercel-sandbox) and access the [open-source SDK and CLI](https://www.npmjs.com/package/sandbox) on npm to get started.

---

### AssistLoop joins the Vercel Agents Marketplace
- Published: 2026-01-30T13:00:00+00:00
- Products: Next.js, Vercel
- Tags: integration, feature, platform
- URL: https://standarddb.com/vercel/assistloop-joins-vercel-agents-marketplace-with-native-next-js-integration--gcmsbryn
- Original: https://vercel.com/changelog/assistloop-joins-the-vercel-agents-marketplace

## New Integration Available

AssistLoop is now available as a native integration in the Vercel Marketplace, enabling developers to add AI-powered customer support directly to their Next.js applications. The integration is designed to work seamlessly within Vercel's ecosystem with minimal setup required.

## Key Capabilities

With AssistLoop, teams can:

- **Install with minimal setup** using a single Agent ID
- **Add AI-powered support** directly to Next.js apps
- **Train custom agents** on internal documentation, FAQs, and knowledge bases
- **Match brand identity** with customizable assistant styling
- **Monitor conversations** and seamlessly hand off to human support when needed

## Getting Started

The integration automatically injects the `NEXT_PUBLIC_ASSISTLOOP_AGENT_ID` environment variable into your Vercel project. Adding the widget to your site requires just a few lines of code using Next.js Script component:

```html
import Script from 'next/script'

<Script
  src="https://assistloop.ai/assistloop-widget.js"
  strategy="afterInteractive"
  onLoad={() => {
    window.AssistLoopWidget?.init({
      agentId: process.env.NEXT_PUBLIC_ASSISTLOOP_AGENT_ID,
    });
  }}
/>
```

Developers can get started immediately by deploying the AssistLoop Next.js template from the Vercel Marketplace. The integration includes unified billing and automatic configuration, eliminating the need for separate dashboards or complex setup processes.

---

### v0.5.76
- Published: 2026-01-29T19:05:12+00:00
- Products: Sim Studio
- Tags: release, bugfix
- URL: https://standarddb.com/sim/sim-studio-v0-5-76-fixes-posthog-session-recording-and-react-hooks-errors--qffcvvvc
- Original: https://github.com/simstudioai/sim/releases/tag/v0.5.76

## Overview

Sim Studio has released v0.5.76, a maintenance update focused on bug fixes and infrastructure improvements. The release addresses key issues in analytics integration and OAuth flows that were affecting user experience.

## Key Changes

**PostHog Analytics Improvements**
- Moved session recording proxy to middleware to support larger payload sizes. This prevents timeout and payload overflow issues when recording extended user sessions with rich interaction data.

**OAuth and React Fixes**
- Fixed a React hooks error in OAuth icon components by properly using `createElement` instead of direct component invocation. This resolves runtime errors when users authenticate through OAuth providers.

**Documentation**
- Updated README documentation to reflect current project status and setup instructions.

## What Developers Need to Know

If you're using Sim Studio, upgrading to v0.5.76 is recommended to avoid potential issues with analytics tracking and OAuth authentication flows. No breaking changes or new features are included in this release, making it a safe upgrade.

The PostHog middleware change is particularly important for applications with longer session recordings or high-frequency tracking events, as it now handles larger payloads without errors.

---

### Time in status
- Published: 2026-01-29T17:55:00+00:00
- Products: Linear
- Tags: feature, integration, platform
- URL: https://standarddb.com/linear/linear-launches-time-in-status-tracking-to-identify-workflow-bottlenecks--myttksnj
- Original: https://linear.app/changelog/2026-01-29-time-in-status

## Time in Status Tracking

Linear has introduced time-in-status metrics across its platform, providing teams with visibility into how long issues linger at each workflow stage. This metric helps identify where work gets blocked or delayed in your development process.

### Key capabilities:

- **Hover previews**: See cumulative time spent in each status by hovering over any issue's status indicator
- **Sorting and filtering**: Order views by `time in status` to surface blocked work, or filter conditions like `In Review > 7 days` to catch stalled issues
- **Analytics integration**: Analyze time-in-status metrics in Insights and Dashboards, slice by time period, and compare performance across your workspace

### Jira Epic Sync

Linear also upgraded its Jira integration to support bi-directional synchronization between Jira Epics and Linear projects. Changes to status, target date, and description now automatically reflect in both applications, eliminating manual sync work.

### Additional Updates

The release includes numerous fixes and improvements:
- Fixed board scroll position restoration when using row grouping
- Improved slash command sub-menus to work with mouse selection
- Enhanced assignee search ranking for project and team members
- Added project creation and editing support on iOS and Android apps
- MCP server updates allowing `assignee` and `lead` fields to be set to null for removal

The time in status display option is available in Linear's views, while the metric is exposed as a new measure in Insights for deeper analysis.

---

### Introducing Moltworker: a self-hosted personal AI agent, minus the minis
- Published: 2026-01-29T14:00:00+00:00
- Products: Cloudflare Workers, Cloudflare
- Tags: release, feature, platform, sdk
- URL: https://standarddb.com/cloudflare/cloudflare-launches-moltworker-serverless-ai-agent-platform-achieves-98-5-npm--ykffrgvn
- Original: https://blog.cloudflare.com/moltworker-self-hosted-ai-agent

## Cloudflare Moltworker: AI Agents Without Dedicated Hardware

Moltworker is a new middleware and set of adapted scripts that allows developers to run Moltbot—an open-source, self-hosted AI personal assistant—on Cloudflare's platform instead of requiring dedicated Mac mini hardware. The solution combines a Worker entry point that acts as an API router with Cloudflare's Sandbox, Browser Rendering, and R2 storage services to create a fully managed AI agent infrastructure.

### How It Works

Moltworker architecture consists of:
- **Entry-point Worker**: Routes API requests and proxies communications between Cloudflare APIs and the isolated environment
- **Sandbox Container**: Runs the standard Moltbot Gateway runtime and its integrations
- **R2 Storage**: Provides persistent object storage
- **AI Gateway**: Manages AI provider integrations with support for Bring Your Own Key (BYOK) and Unified Billing
- **Cloudflare Access**: Secures the entire stack with authentication and authorization

### Node.js Compatibility Milestone

This announcement highlights a significant achievement: **Cloudflare Workers now achieves 98.5% compatibility with the top 1,000 npm packages**. An internal experiment testing the most popular npm packages in Workers found that only 15 packages genuinely failed to work (excluding build tools, CLI tools, and browser-only packages). This improvement stems from native support for Node.js APIs, including `node:fs`, reducing reliance on workarounds like memfs and enabling easier maintenance of critical dependencies like Playwright.

### Developer Impact

Developers can now run complex AI applications directly on Cloudflare's global network without provisioning separate hardware. The platform automatically scales and benefits from Cloudflare's security and performance infrastructure. Integration with popular AI providers through AI Gateway and support for Bring Your Own Key secrets management simplifies credential handling.

**Next steps**: Developers interested in running Moltbot on Cloudflare should visit the [Moltworker GitHub repository](https://github.com/cloudflare/moltworker) for setup instructions and deployment guides.

---

### v0.5.75
- Published: 2026-01-29T07:10:21+00:00
- Products: Sim Studio
- Tags: release, feature, api, bugfix, platform
- URL: https://standarddb.com/sim/sim-releases-v0-5-75-with-nested-workflow-execution-snapshots-and-youtube-api--scrpgpyy
- Original: https://github.com/simstudioai/sim/releases/tag/v0.5.75

## New Features

**Child Workflow Snapshots**: The release adds nested execution snapshots for child workflows, enabling better visibility and debugging of workflow execution across multiple levels.

**Enhanced YouTube Integration**: YouTube tool support has been expanded significantly with new capabilities:
- Captions extraction
- Trending video discovery
- Video categories support
- Improved overall API coverage

**Execution and Configuration Improvements**:
- **API Block Timeout Configuration**: New ability to configure timeout settings for API blocks
- **Structured Terminal Output**: Terminal output is now structured for better parsing and readability
- **Deployment Version Descriptions**: Added support for descriptions of deployment versions
- **Run From/Until Block Executor**: New functionality to execute workflows starting from or stopping at specific blocks

## Improvements & Fixes

The release includes refinements to the UI/UX layer:
- Workflow preview breadcrumbs now include current workflow badges
- Enhanced trigger input sanitization for improved data quality
- Refactored search modal with quick navigation items and improved command palette integration
- Icon updates for consistency in the action bar

Several important bug fixes address workflow execution, logging, and platform stability:
- Fixed logs workspace delivery type issues
- Corrected copilot panning behavior on workflows
- Resolved terminal spacing and subflow preview issues
- Fixed trace span accuracy in child workflow error logs
- Fixed loops behavior when operating on empty collections
- Updated Helm configuration for cert-manager compatibility

---

### Retiring GPT-4o, GPT-4.1, GPT-4.1 mini, and OpenAI o4-mini in ChatGPT
- Published: 2026-01-29T00:00:00+00:00
- Products: ChatGPT, OpenAI
- Tags: deprecation, release, model
- URL: https://standarddb.com/openai/openai-retires-gpt-4o-from-chatgpt-shifts-users-to-gpt-5-2-with-enhanced--bgqfjrkb
- Original: https://openai.com/index/retiring-gpt-4o-and-older-models

## Model Retirement Timeline

OpenAI will retire four models from ChatGPT effective **February 13, 2026**:
- GPT-4o
- GPT-4.1
- GPT-4.1 mini
- o4-mini

Note that these changes apply to ChatGPT only; the OpenAI API has no changes at this time.

## Why GPT-4o Is Being Retired

While multiple older models are being deprecated, GPT-4o deserves special context. OpenAI initially deprecated the model but restored it after hearing feedback from Plus and Pro users who needed more time to transition key workflows—particularly creative ideation—and preferred its conversational style and warmth.

That user feedback directly informed the development of GPT-5.1 and GPT-5.2, which now include:
- **Enhanced personality and creativity** improvements
- **Customizable response styles** including base personalities (e.g., "Friendly") and granular controls for warmth and enthusiasm
- Better support for the creative use cases GPT-4o users valued

## User Adoption Has Shifted

OpenAI is announcing the retirement today because the company's goals have been met: vast majority usage has migrated to newer models, with only 0.1% of daily active ChatGPT users still selecting GPT-4o. This makes the retirement decision more defensible and allows the team to focus development efforts on widely-used models.

## Broader Product Direction

Beyond model retirement, OpenAI is advancing ChatGPT in several areas users requested:
- Further personality and creativity improvements
- Reducing unnecessary refusals and overly cautious responses
- Developing an 18+ version grounded in treating adults like adults
- Rolling out age prediction for users under 18 in most markets

The company acknowledges this change will frustrate some users and commits to clear communication about model retirements and timelines.

---

### Your logs' final destination is in GA. You always end up here anyway
- Published: 2026-01-29T00:00:00+00:00
- Products: PostHog
- Tags: release, feature, platform, integration
- URL: https://standarddb.com/posthog/posthog-s-logs-feature-exits-beta-with-ga-release-unified-debugging-platform--yqgwsrlm
- Original: https://posthog.com/blog/logs-ga

## Logs Now Generally Available

PostHog has announced general availability of its **Logs** feature, moving it from beta status. This release positions Logs as the final destination for developers investigating system issues—whether starting from an alert, error, or user report, debugging workflows ultimately converge here.

## Key Capabilities and Integration

**Logs** is built on standard OpenTelemetry ingestion (OTLP), eliminating the need for proprietary SDKs or new instrumentation models. Developers can route existing OpenTelemetry logs with minimal configuration. The feature integrates seamlessly with PostHog's broader platform:

- Frontend and backend logs live together and are automatically linked to users and sessions
- Browser logs captured via PostHog JS are ingested alongside backend logs
- Logs tie directly to session replays and error tracking, creating a unified investigation flow
- Structured filtering by service, severity, user IDs, and custom attributes
- AI-powered insights can summarize patterns and flag anomalies

## Pricing and Availability

The general availability release includes:

- **Free tier**: 50 GB/month with 7-day retention
- **Paid tier**: $0.25 per GiB ingested (volume discount to $0.15/GiB at 300+ GB)
- No per-seat or query fees
- All beta users receive one complimentary month

The release was informed by PostHog's internal dogfooding—engineers used Logs to debug PostHog itself, uncovering cases where previous logging silently hid critical issues like repeated out-of-memory crashes caused by unintended large payload logging.