← Back
LangChain
LangChain releases skills for AI coding agents, boosting Claude Code performance from 29% to 95%
· featuresdkrelease · blog.langchain.com ↗

What are Skills?

Skills are curated instructions, scripts, and resources designed to improve coding agent performance in specialized domains. Rather than loading all available tools at once (which historically degrades agent performance), skills are dynamically retrieved through progressive disclosure—agents only access a skill when relevant to the current task.

LangChain Skills Release

LangChain has released an initial set of 11 skills across three categories:

  • LangChain: Guidance on create_agent(), middleware, and tool patterns for classic agent loops
  • LangGraph: Documentation on primitives, Human-in-the-Loop support, and durable execution
  • DeepAgents: Resources for the open-source Deep Agents package, including prebuilt middleware and filesystem integration

The skills are portable, shareable markdown files available in the langchain-skills GitHub repository and can be integrated into any coding agent supporting skill functionality.

Performance Impact

Testing with Claude Code and Sonnet 4.6 demonstrated significant improvements:

Scenario Pass Rate
Without Skills 25%
With Skills 95%

Getting Started

Install skills locally or globally using the npx skills command:

# Local installation
npx skills add langchain-ai/langchain-skills --skill '*' --yes

# Global installation
npx skills add langchain-ai/langchain-skills --skill '*' --yes --global

# Link to Claude Code specifically
npx skills add langchain-ai/langchain-skills --agent claude-code --skill '*' --yes --global

LangChain plans to expand the skills library as new capabilities are added to their open-source ecosystem and LangSmith platform. They also released a companion set of LangSmith skills in the same release.