github-md-mcp

bPol/github-md-mcp

3.1

If you are the rightful owner of github-md-mcp and would like to certify it and/or have it hosted online, please leave a comment on the right or send an email to henry@mcphub.com.

A serverless MCP server that exposes Markdown files from GitHub repositories as AI-ready resources.

github-md-mcp

A serverless MCP (Model Context Protocol) server that exposes Markdown files from any public or private GitHub repository as AI-ready resources.

Features

  • ✅ List all .md files in a repo
  • 📄 Get file content by path or by MCP ID (mcp://files/...)
  • 🔎 Search filenames and Markdown content for keywords
  • 🔐 Optional API-key and HMAC authentication
  • ⚡ Runs as an AWS Lambda (Node 20+) or any Node.js serverless runtime

Why

Connecting large-language-model agents to a knowledge base usually requires custom plumbing.
This Lambda acts as a bridge: your GitHub repo becomes an MCP server the agent can query in real time.

Quick start

1. Deploy

Clone and deploy the index.mjs as an AWS Lambda (Node.js 20+) or any environment that accepts an HTTP POST endpoint.

Set these environment variables:

VariableDescription
GH_OWNERGitHub user or org name
GH_REPORepository name
GH_REF(optional) Branch, defaults to repo default branch
GITHUB_TOKENGitHub Personal Access Token with repo (read)
MCP_API_KEY(optional) Bearer token required for requests
MCP_HMAC_SECRET(optional) HMAC secret to sign requests
ALLOWED_ORIGINS(optional) Comma-separated list for CORS

2. MCP client configuration

Point your MCP client (for example an AI agent) to the Lambda URL:

  • initialize → handshake
  • tools/list → discover tools
  • tools/call with:
{
  "jsonrpc": "2.0",
  "id": "1",
  "method": "tools/call",
  "params": {
    "name": "get_md_file",
    "arguments": {
      "path": "mcp://files/todo-current.md"
    }
  }
}