Web Search and Web Fetch for OpenClaw

Ollama now ships with a web search and fetch plugin for OpenClaw, allowing both local and cloud-based models to search the web for latest content and news. The web fetch feature can extract readable content from web pages for model processing, though it does not execute JavaScript.

When using local models with web search in OpenClaw, you must sign in with ollama signin. To launch OpenClaw with these capabilities:

ollama launch openclaw

If you already have OpenClaw configured, install the web search plugin directly:

openclaw plugins install @ollama/openclaw-web-search

Non-Interactive (Headless) Mode

The ollama launch command now supports non-interactive mode, ideal for automation scenarios:

• Docker/Containers: Spin up integrations as pipeline steps to run evaluations, test prompts, or validate model behavior during builds, then tear down automatically
• CI/CD: Generate code reviews, security checks, and other automated tasks within your CI pipeline
• Scripts/Automation: Kick off automated tasks without requiring user interaction

Requirements and Usage

• The --model flag must be specified to run in headless mode
• Use the --yes flag to automatically pull the model and skip interactive selectors

Example command:

ollama launch claude --model kimi-k2.5:cloud --yes -- -p "how does this repository work?"

Additional Changes

• ollama launch openclaw now uses Ollama's official authentication and model provider
• Improvements to Ollama's benchmarking tool
• Fixed --install-daemon behavior when systemd is unavailable