github/github-mcp-server

What it solves

This project provides a bridge between AI agents and the GitHub platform. It allows AI tools (like chatbots, assistants, and agents) to interact with GitHub repositories, issues, pull requests, and workflows using natural language, removing the need for manual API calls or custom integrations for every tool.

How it works

It implements the Model Context Protocol (MCP), which standardizes how AI models can access external tools and data. The server can be deployed either as a remote hosted service (managed by GitHub) or as a local instance (via Docker or a Go binary). It authenticates via OAuth or Personal Access Tokens (PAT) to securely access the user's GitHub account and provides a set of "toolsets" that the AI can call to perform actions like reading code, managing issues, or monitoring CI/CD pipelines.

Who it’s for

Developers and teams who use AI-powered IDEs (like VS Code, Cursor, Windsurf) or AI agents (like Claude Desktop) and want those tools to have direct, authenticated access to their GitHub codebase and project management tools.

Highlights

• Multi-Host Support: Compatible with a wide range of AI hosts including VS Code, Claude Desktop, Cursor, Windsurf, and Zed.
• Flexible Deployment: Offers both a remote server for quick setup and a local server for more control.
• Enterprise Ready: Supports GitHub Enterprise Cloud and GitHub Enterprise Server.
• Granular Control: Allows enabling or disabling specific toolsets to optimize LLM tool selection and reduce context window usage.
• Comprehensive GitHub Integration: Covers everything from repository browsing and code analysis to issue triage and GitHub Actions monitoring.