Git PR MCP Server
An MCP server that enables Git repository operations and GitHub PR workflows, allowing users to manage repositories, create branches, commit changes, and create pull requests through natural language.
README
Git PR MCP Server
A Model Context Protocol (MCP) server for Git and Pull Request operations, providing tools for repository management and PR workflows.
Built with FastMCP and runs in SSE mode by default on 0.0.0.0:8000.
Features
This MCP server provides the following tools and capabilities:
Core Git Operations (on specified repository path):
- get_git_status: Get the current git status of a repository.
- list_branches: List all branches in the repository (with optional remote branches).
- create_pr_summary: Create a summary for a pull request based on git diff.
- get_commit_history: Get commit history for a branch.
- get_git_diff: Get git diff between commits, branches, or working directory.
Automated PR Workflow (operates on an internally managed "active" repository):
- clone_repository: Clones a GitHub repository into a managed temporary directory, making it the "active" repository for subsequent operations. Automatically cleans up any previously active repository's temporary directory.
- create_git_branch: Creates a new branch in the active repository.
- write_file_in_repo: Creates or overwrites files within the active repository.
- git_commit_changes: Stages all changes (
git add .) and commits them in the active repository. - git_push_branch: Pushes a specified branch from the active repository to its remote origin.
- create_github_pr: Creates a pull request on GitHub for the active repository using the PyGithub library. Requires
GITHUB_TOKEN.
State Management:
- The server maintains the state of the currently "active" cloned repository (path, URL, owner, name) in an
active_repo_state.jsonfile. This allows some persistence across server restarts, though the temporary clone directory itself might be OS-managed.
Installation
- Clone this repository:
git clone <repository-url>
cd git-pr-mcp
- Install using uv:
uv sync
Usage
The server runs in SSE mode by default on 0.0.0.0:9999 and is configured via environment variables. Here's a sample .env file:
FASTMCP_HOST=0.0.0.0
FASTMCP_PORT=9999
GITHUB_TOKEN=your_github_token
Quick Start
# Run with defaults (0.0.0.0:8000)
uv run python main.py
# Or run the server module directly
uv run python -m src.git_pr_mcp.server
You can then access the MCP server at http://localhost:9999/sse.
推荐服务器
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。
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。
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。
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。
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。
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。
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。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
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。