GitHub MCP Server
Enables interacting with GitHub repositories for pull requests, code, and reviews through natural language.
README
GitHub MCP Server
FastMCP server for GitHub pull requests, code, and reviews.
Features
- Repository/account helpers: authenticated user, repository metadata, branches, branch creation, ref comparison.
- Pull requests: list, get, create, update, merge, changed files, commits, raw diff.
- Code: read files, list directories, search code, create/update text files, delete files.
- Reviews: list reviews, list review comments, list conversation comments, add conversation comments, create/submit/dismiss reviews, add file-level review comments, request reviewers.
Create .env file
GITHUB_TOKEN=github_pat_or_personal_access_token_here
GITHUB_OWNER=owner-or-org
GITHUB_REPO=repo-name
# Optional
GITHUB_API_URL=https://api.github.com
GITHUB_API_VERSION=2022-11-28
GITHUB_OWNER and GITHUB_REPO are optional if every tool call passes owner and repo.
For fine-grained GitHub tokens, grant only the permissions you need. Common permissions:
- Pull requests: read/write
- Contents: read/write
- Metadata: read
- Issues: read/write, for pull request conversation comments
Create a virtual environment
cd PROJECT_FOLDER
python3 -m venv .venv
Activate it
source .venv/bin/activate
Install dependencies
pip install fastmcp requests python-dotenv
Freeze the requirements
pip freeze > requirements.txt
Test it
python3 server.py
You should see:
Starting MCP server 'github-mcp-server' with transport 'stdio'
Configure OpenCode
{
"$schema": "https://opencode.ai/config.json",
"mcp": {
"github": {
"type": "local",
"command": [
"PATH_TO_YOUR_MCP_PROJECT/.venv/bin/python",
"PATH_TO_YOUR_MCP_PROJECT/server.py"
],
"enabled": true
}
}
}
Docker / HTTP mode
docker compose up --build
HTTP mode listens on port 8002.
Useful Next Features
- Issues and labels: create issues, triage labels, milestone assignment, linked PR discovery.
- Checks and CI: list check suites/runs, rerun failed checks, summarize failing annotations.
- Actions workflows: list workflow runs, trigger
workflow_dispatch, download artifacts. - Releases and tags: draft releases, generate changelog notes, compare release refs.
- Branch protection: inspect required checks, review rules, and merge restrictions.
- Security: Dependabot alerts, secret scanning alerts, code scanning alerts, SARIF summaries.
- Repository admin: collaborators, teams, permissions, rulesets, webhooks.
- Notifications: list assigned/review-requested items across repositories.
- Project planning: GitHub Projects items, statuses, and field updates.
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。