github-mcp
Local MCP server for safe GitHub developer workflows. Enables pulling main, creating feature branches, pushing them, and opening ready-for-review pull requests via tools like github_create_feature_branch and github_create_pull_request.
README
GitHub MCP Server
Local OpenCode MCP server for safe GitHub developer workflows: pull main, create a feature branch, push it, and open a ready-for-review pull request.
Environment
Set these variables in your shell. Do not commit real secrets.
export GH_TOKEN="your_github_token_here"
export GITHUB_MCP_ALLOWED_ROOTS="/Users/matiasbinagora/Projects"
export GITHUB_MCP_DEFAULT_BASE_BRANCH="main"
export GITHUB_MCP_DEFAULT_REMOTE="origin"
export GITHUB_MCP_REQUIRE_CLEAN_WORKTREE="true"
GH_TOKEN is recommended. If it is not set, the server can still work when gh is already authenticated locally.
Build
npm install
npm run build
Smoke Test
This creates a temporary local bare repo, pulls main, creates a test branch, commits a file, pushes the branch, and verifies the remote branch exists.
npm run smoke
GitHub Integration Test
This creates a real branch and ready-for-review PR in the configured test repo.
export GITHUB_MCP_SMOKE_REPO_PATH="/Users/matiasbinagora/Projects/binagora"
npm run smoke:github
OpenCode Config
Add this to the global OpenCode config at ~/.config/opencode/opencode.json.
{
"$schema": "https://opencode.ai/config.json",
"mcp": {
"github": {
"type": "local",
"command": ["node", "/Users/matiasbinagora/Projects/github-mcp/dist/index.js"],
"enabled": true,
"environment": {
"GH_TOKEN": "{env:GH_TOKEN}",
"GITHUB_MCP_ALLOWED_ROOTS": "/Users/matiasbinagora/Projects",
"GITHUB_MCP_DEFAULT_BASE_BRANCH": "main",
"GITHUB_MCP_DEFAULT_REMOTE": "origin",
"GITHUB_MCP_REQUIRE_CLEAN_WORKTREE": "true"
}
}
}
}
If OpenCode is launched before your shell exports GH_TOKEN, use this command instead so the MCP server sources ~/.zshrc before starting:
"command": [
"zsh",
"-lc",
"source \"$HOME/.zshrc\" >/dev/null 2>&1; exec node \"/Users/matiasbinagora/Projects/github-mcp/dist/index.js\""
]
Tools
github_repo_statusgithub_pull_maingithub_create_feature_branchgithub_push_feature_branchgithub_create_pull_requestgithub_prepare_feature_workflow
Safety Notes
- Repositories must be inside
GITHUB_MCP_ALLOWED_ROOTS. - No generic shell execution tool is exposed.
- Direct pushes to
mainormasterare refused. - Force pushes and branch deletion are not implemented.
- A clean worktree is required before pulling, branch creation, and PR creation by default.
- Branch names are validated with
git check-ref-format --branch.
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。
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。
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。
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。
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。
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。
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。
mcp-server-qdrant
这个仓库展示了如何为向量搜索引擎 Qdrant 创建一个 MCP (Managed Control Plane) 服务器的示例。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。