GitMCP MCP Server

What is GitMCP?

GitMCP transforms any GitHub project — repositories or GitHub Pages — into an accessible documentation and code hub for AI-powered tools. By acting as an MCP server, it bridges your AI assistants (such as Cursor, Claude Desktop, Windsurf, VSCode, and others) to the latest source code and documentation, allowing them to answer questions, provide accurate code examples, and eliminate the risk of outdated or hallucinated responses. GitMCP is cloud-based, zero-setup, private, and works with any AI assistant supporting MCP.

How to Configure GitMCP

1. Select your target:

   • For a specific GitHub repository: https://gitmcp.io/{owner}/{repo}
   • For a GitHub Pages site: https://{owner}.gitmcp.io/{repo}
   • For dynamic access to any repo: https://gitmcp.io/docs
   • Replace {owner} and {repo} with your chosen values.

2. Configure your AI tool:

   • Cursor: Edit ~/.cursor/mcp.json and add:

     {
       "mcpServers": {
         "gitmcp": {
           "url": "https://gitmcp.io/{owner}/{repo}"
         }
       }
     }
     

   • Claude Desktop: In Settings > Developer > Edit Config, add:

     {
       "mcpServers": {
         "gitmcp": {
           "command": "npx",
           "args": [
             "mcp-remote",
             "https://gitmcp.io/{owner}/{repo}"
           ]
         }
       }
     }
     

   • Windsurf: Edit ~/.codeium/windsurf/mcp_config.json as:

     {
       "mcpServers": {
         "gitmcp": {
           "serverUrl": "https://gitmcp.io/{owner}/{repo}"
         }
       }
     }
     

   • VSCode: Edit .vscode/mcp.json in your project:

     {
       "servers": {
         "gitmcp": {
           "type": "sse",
           "url": "https://gitmcp.io/{owner}/{repo}"
         }
       }
     }
     

   • Highlight AI: Add a custom SSE plugin with:

     Name: gitmcp
     SSE URL: https://gitmcp.io/{owner}/{repo}
     

   See official docs for more integrations and details.

How to Use GitMCP

1. Connect GitMCP to your AI tool using the appropriate configuration (see above).
2. Start your AI assistant and choose or specify the GitMCP server URL for the target repository or site.
3. Prompt your assistant with questions about project documentation, API usage, implementation details, or code searching.
4. The AI will seamlessly invoke GitMCP tools to fetch documentation, search code or docs, or retrieve referenced content.
5. Review the improved, grounded responses provided by your AI, with full traceability to source material.

Example prompts:

• "How do I use the Playwright MCP?" (AI obtains latest docs)
• "What is the usage of this class in repo X?" (AI searches code)
• "Show me example API calls for library Y." (AI fetches precise docs/examples)

Key Features

• Direct, real-time access to any GitHub project's documentation and code, always up to date
• Zero installation and configuration overhead: Fully cloud-based, needs only a URL
• Reduces model hallucinations by grounding answers in actual source code and docs
• Supports specific and generic access: Pin to a repo/site or use on-demand for any repository
• Smart search tools for both documentation and code, with minimal token footprint
• Works with popular AI assistants and IDEs with simple configuration
• Privacy-focused: No user data or queries stored; open-source for transparency and self-hosting
• Embeddable chat and repo badges for enhanced developer experience and documentation discoverability

Use Cases

• Developers integrating new or niche open-source libraries, needing the latest usage patterns or documentation examples not yet covered by LLM training data.
• AI code assistants providing code completion, answers, or automated documentation lookups for unfamiliar repos.
• Teams working on private forks or rapid-release codebases, who want their internal documentation and code always available to their AI tooling.
• Reducing AI errors and hallucinations by ensuring every answer is based on the current state of the codebase and public docs.
• Providing instant, accurate onboarding support for newcomers, with live documentation search/chat powered by GitMCP.

FAQ

Q1: Does GitMCP work with any AI assistant? Yes, GitMCP is compatible with any assistant supporting the Model Context Protocol, such as Cursor, Claude Desktop, Windsurf, VSCode, Highlight AI, and more.

Q2: Does GitMCP support all GitHub repositories? Absolutely! GitMCP works with any public GitHub repository or Pages site. It automatically picks up the llms.txt if available, or falls back to README.md and other main docs.

Q3: Is GitMCP free to use? Yes, GitMCP is a completely free and open-source community service, with no registration or usage costs.

Q4: What about privacy — does it store my queries or personal data? No. GitMCP does not require login, does not collect or persist any personal data or queries, and can be self-hosted for complete control. Only publicly available contents are accessed, and only when explicitly requested.

Q5: How can I add GitMCP support or badge to my own project? Simply add the official badge to your README as described in the documentation, or set up your own instance for internal work.