GitHub MCP

GitHub MCP

Enables self-hosted, read-only remote interaction with GitHub repositories, allowing listing repositories, searching code, and inspecting commits, pull requests, issues, and diffs via authenticated MCP clients.

Category
访问服务器

README

GitHub MCP

A generic, read-only remote MCP connector for querying GitHub repositories selected through a GitHub App installation. It is designed for self-hosting and contains no GitHub, Cloudflare, or MCP-client account identifiers.

The connector uses two authorization layers:

  1. A user authorizes the GitHub App to prove their GitHub identity. Only operator-configured logins may finish connecting.
  2. Tool calls use short-lived GitHub App installation tokens restricted to the requested installed repository.

The GitHub user token is discarded after identity verification. The connector exposes no write tools.

Capabilities

  • Discover repositories selected in GitHub App installations.
  • Search code and retrieve text files or directory listings.
  • Inspect commits, pull requests, diffs, reviews, and comments.
  • Inspect issues and issue comments.
  • Authenticate remote MCP clients using OAuth, PKCE, and Dynamic Client Registration.

Local development

Requirements: Node 22 or newer and Yarn 1.22.

yarn install
yarn type-check
yarn lint
yarn test
yarn build

Tests use fake identities and mocked GitHub services. No GitHub App or Cloudflare account is needed for local verification.

To run the Worker against real services, follow the operator setup guide. Do not commit wrangler.jsonc, local secrets, or GitHub App private keys.

MCP tools

Tool Purpose
list_repositories List repositories selected in the app's installations.
search_code Search within one installed repository.
list_directory List one repository directory at a ref.
get_file Retrieve a UTF-8 text file, optionally by line range.
list_commits / get_commit Inspect repository history.
list_pull_requests / get_pull_request Inspect pull requests, files, reviews, and comments.
get_pull_request_diff Retrieve a pull request's unified diff.
list_issues / get_issue Inspect issues and comments.

Results are live GitHub API responses, normalized into bounded JSON with canonical GitHub URLs. Pagination defaults to 20 items and caps at 50. Text and diff output caps at 100,000 characters.

Security model

  • AUTHORIZED_GITHUB_LOGINS is mandatory and fails closed when empty.
  • GitHub App installations determine which repositories exist to the connector.
  • Each repository tool call mints a one-hour token restricted to that repository.
  • Required GitHub App permissions are read-only Metadata, Contents, Issues, and Pull Requests.
  • OAuth codes, user tokens, installation tokens, client secrets, and private keys are never returned or intentionally logged.

This project is intended as a personal or small-team connector, not a hosted multi-tenant service.

License

This project is available under the MIT License.

推荐服务器

Baidu Map

Baidu Map

百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。

官方
精选
JavaScript
Playwright MCP Server

Playwright MCP Server

一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。

官方
精选
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。

官方
精选
本地
TypeScript
VeyraX

VeyraX

一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。

官方
精选
本地
graphlit-mcp-server

graphlit-mcp-server

模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。

官方
精选
TypeScript
Kagi MCP Server

Kagi MCP Server

一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

Exa MCP Server

模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。

官方
精选