GitHub MCP Server

Minimal Model Context Protocol (MCP) server that exposes GitHub-focused tools for AI assistants.

🔧 Quick Start (Local)

# 1. Clone & install git clone <repo-url> cd github-review-mcp pnpm install # 2. Add credentials echo "GITHUB_TOKEN=ghp_your_token_here" > .env # 3. Run the server pnpm start

The server listens on the port specified by your MCP client (default 3000).

Related MCP server: Azure Revisor MCP Server

🖥️ Using from an MCP Client

Cursor IDE

Add this to Settings → Extensions → MCP (or your mcp-servers.json):

{ "mcpServers": { "github-review-server": { "command": "node", "args": ["/absolute/path/github-review-mcp/src/index.js"], "env": { "GITHUB_TOKEN": "ghp_your_token_here" } } } }

Generic MCP Client

Point the client to node /absolute/path/github-review-mcp/src/index.js and pass GITHUB_TOKEN in the environment.

🐳 Docker (Hosted)

A Dockerfile mirroring the hosted image is included.

# Build docker build -t gh-mcp . # Run docker run -e GITHUB_TOKEN=ghp_your_token_here -p 3000:3000 gh-mcp

Attach the container to your MCP client using http://localhost:3000 (or the mapped port) as the command.

⚙️ Environment Variables

Variable	Description	Default
`GITHUB_TOKEN`	GitHub Personal Access Token (required)	-
`MAX_PATCH_SIZE`	Maximum diff patch size (chars)	2000
`MAX_FILES_TO_REVIEW`	Maximum files processed per PR	50
`REQUEST_TIMEOUT`	HTTP request timeout (ms)	30000
`LOG_LEVEL`	Logging level ( `debug` , `info` , …)	info
`ENABLE_DEBUG`	Verbose logging ( `true` / `false` )	false