UV Documentation MCP Server

What is UV Documentation MCP Server?

An MCP server that provides programmatic access to UV documentation through a standardized API, exposing high‑level documentation sections and detailed command documentation (including subsections) as individual MCP resources.

How to use UV Documentation MCP Server?

Requires Python 3.13+ (minimum 3.10 for MCP SDK compatibility). Clone the repository, create a virtual environment, install dependencies with pip install -e ., then run the server with uv run uv-docs. For use with Claude Desktop, configure the MCP server settings using the uv command with the appropriate --directory and run arguments.

Key features of UV Documentation MCP Server

• Hierarchical resource structure using uv-docs:// URIs (sections, commands, subsections)
• Tools: update_cache and search_documentation for cache and real‑time queries
• Prompt template summarize-docs for AI documentation summaries
• Version‑aware caching with segmented section storage
• Fully open‑source and publishable to PyPI

Use cases of UV Documentation MCP Server

• AI assistants retrieving precise UV documentation sections on demand
• Developers automating the lookup of UV CLI commands and their options
• Integrating UV documentation into custom MCP‑compatible tools or workflows
• Generating summaries of selected UV documentation areas for reporting

FAQ from UV Documentation MCP Server

What does the resource layer provide?

Resources expose UV documentation via URIs like uv-docs://cli, uv-docs://cli/uv-cache, and uv-docs://cli/uv-cache/uv-cache-dir, returning JSON with metadata and full documentation content.

How can I search documentation in real time?

Use the search_documentation tool, which accepts a query string and performs a live search over the UV documentation.

What are the runtime dependencies?

Python 3.10+ (3.13+ recommended) and the MCP Python SDK. The project is installed via pip and run with uv (or uvx for published versions).

How does caching work?

A version‑aware cache is initialized at startup. Each documentation section (CLI, settings, resolver) is cached separately. The cache is automatically updated when the live version changes, or manually via the update_cache tool.

Is authentication or specific transport required?

No authentication or special transport is mentioned. The server uses standard MCP communication over stdio, configured through MCP client settings.