GitHub

The GitHub MCP Server connects AI tools directly to GitHub's platform. It lets AI agents, assistants and chatbots read repositories and code files, manage issues and pull requests, analyse code and automate workflows, all through natural language.

The page describes several uses. Developers can browse and query code, search files, analyse commits and understand project structure across any repository they have access to. They can have AI help triage bugs, review code changes and maintain project boards. The server can also monitor GitHub Actions workflow runs, analyse build failures and manage releases. Security findings, Dependabot alerts and code patterns can be reviewed through it, and it can access discussions and notifications for a team.

The server comes in two forms. The remote version is hosted by GitHub at a Copilot API endpoint and is the simplest route. It works with MCP hosts that support remote servers, including VS Code 1.101 or later, Claude Desktop, Cursor and Windsurf, and the page lists installation guides for several other hosts. The remote server supports OAuth or a personal access token. The local version runs in a Docker container from ghcr.io/github/github-mcp-server. On github.com, the local server can log in with OAuth on first use, or it can use a GITHUB_PERSONAL_ACCESS_TOKEN environment variable instead.

The page also covers GitHub Enterprise Server and GitHub Enterprise Cloud with data residency, which are configured with a host setting. It recommends storing tokens in environment variables, granting only the scopes needed, and rotating tokens regularly.

The server is open source under the MIT licence and is written in Go. The Go module can be imported as a library. The repository shows 34.4k stars and 5.3k forks.

The page does not give a price for the server itself. GitHub's pricing page lists a free plan and paid plans for teams and enterprises, but the server is not listed as a separate product on it.