MongoDB MCP Server for Vercel

A read-only MongoDB MCP (Model Context Protocol) server designed for serverless deployment on Vercel. Provides secure, limited access to MongoDB databases for AI assistants and MCP-compatible clients.

Features

• 🔒 Secure by default - API key authentication, read-only operations, query limits

• ⚡ Serverless optimized - Designed for Vercel's edge/serverless environment

• 🛡️ Safety guardrails - Dangerous operators blocked, query timeouts, result limits

• 📊 6 MongoDB tools - find, aggregate, count, list-collections, explain, collection-schema

Related MCP server: MongoDB MCP Server by CData

Quick Start

1. Install dependencies

npm install

2. Configure environment variables

Create a .env.local file:

# MongoDB Connection (use a read-only user!)
MONGODB_URI=mongodb+srv://readonly_user:password@cluster.mongodb.net
MONGODB_DB=your_database_name

# API Key for authentication (generate with: openssl rand -base64 32)
API_KEY=your_generated_api_key

3. Run locally

npm run dev

The MCP server will be available at http://localhost:3000/mcp

Tools

Security

Built-in protections

• Fixed database - Only the database specified in MONGODB_DB is accessible

• API key required - All requests must include X-API-Key header

• Query limits - Max 100 documents per query, 30s timeout

• Blocked operators - $where, $function, $accumulator are rejected

• Blocked stages - $out, $merge, $lookup are rejected in aggregations

• EJSON serialization - Proper handling of BSON types (ObjectId, Date, etc.)

Recommendations

1. Use a read-only MongoDB user - Create a user with only read role

2. Rotate API keys - Generate new keys periodically

3. Monitor usage - Enable Vercel Analytics to track requests

Usage

With cURL

# Initialize connection
curl -X POST http://localhost:3000/mcp \
  -H "Content-Type: application/json" \
  -H "Accept: application/json, text/event-stream" \
  -H "X-API-Key: YOUR_API_KEY" \
  -d '{"jsonrpc":"2.0","method":"initialize","params":{"protocolVersion":"2024-11-05","capabilities":{},"clientInfo":{"name":"test","version":"1.0"}},"id":1}'

# List available tools
curl -X POST http://localhost:3000/mcp \
  -H "Content-Type: application/json" \
  -H "Accept: application/json, text/event-stream" \
  -H "X-API-Key: YOUR_API_KEY" \
  -d '{"jsonrpc":"2.0","method":"tools/list","params":{},"id":2}'

# Call a tool (list collections)
curl -X POST http://localhost:3000/mcp \
  -H "Content-Type: application/json" \
  -H "Accept: application/json, text/event-stream" \
  -H "X-API-Key: YOUR_API_KEY" \
  -d '{"jsonrpc":"2.0","method":"tools/call","params":{"name":"list-collections","arguments":{}},"id":3}'

# Find documents
curl -X POST http://localhost:3000/mcp \
  -H "Content-Type: application/json" \
  -H "Accept: application/json, text/event-stream" \
  -H "X-API-Key: YOUR_API_KEY" \
  -d '{"jsonrpc":"2.0","method":"tools/call","params":{"name":"find","arguments":{"collection":"users","filter":{"status":"active"},"limit":10}},"id":4}'

With MCP Inspector

1. Open MCP Inspector

2. Set Transport Type: Streamable HTTP

3. Set URL: http://localhost:3000/mcp

4. In Authentication > Custom Headers, add:

   • X-API-Key: your API key

   • Accept: application/json, text/event-stream

5. Enable both header toggles and click Connect

With Cursor

Add to your .cursor/mcp.json:

{
  "mcpServers": {
    "mongodb": {
      "url": "http://localhost:3000/mcp",
      "headers": {
        "X-API-Key": "YOUR_API_KEY"
      }
    }
  }
}

For production deployment:

{
  "mcpServers": {
    "mongodb": {
      "url": "https://your-app.vercel.app/mcp",
      "headers": {
        "X-API-Key": "YOUR_API_KEY"
      }
    }
  }
}

Deploy to Vercel

1. Push to GitHub

git init
git add .
git commit -m "Initial commit"
git remote add origin https://github.com/your-username/your-repo.git
git push -u origin main

2. Import to Vercel

1. Go to vercel.com/new

2. Import your GitHub repository

3. Add environment variables:

   • MONGODB_URI

   • MONGODB_DB

   • API_KEY

4. Deploy

3. Configure MongoDB Network Access

Make sure your MongoDB Atlas cluster allows connections from Vercel:

• Add 0.0.0.0/0 to IP Access List (for serverless), or

• Use Vercel's static IP addresses

API Reference

Tool: find

Query documents from a collection.

{
  "collection": "users",
  "filter": { "status": "active" },
  "projection": { "name": 1, "email": 1, "_id": 0 },
  "sort": { "createdAt": -1 },
  "limit": 10
}

Tool: aggregate

Run an aggregation pipeline.

{
  "collection": "orders",
  "pipeline": [
    { "$match": { "status": "completed" } },
    { "$group": { "_id": "$customerId", "total": { "$sum": "$amount" } } },
    { "$sort": { "total": -1 } }
  ]
}

Tool: count

Count documents matching a filter.

{
  "collection": "products",
  "filter": { "inStock": true }
}

Tool: list-collections

List all collections in the database. No arguments required.

{}

Tool: explain

Get the execution plan for a query.

{
  "collection": "users",
  "operation": "find",
  "operationArgs": {
    "filter": { "email": "test@example.com" }
  }
}

Tool: collection-schema

Infer schema by sampling documents.

{
  "collection": "users",
  "sampleSize": 100
}

Environment Variables

Based On

This project combines ideas and code from:

• vercel-labs/mcp-for-next.js - Vercel's official template for deploying MCP servers on Next.js. Provides the serverless-compatible architecture and mcp-handler integration.

• mongodb-js/mongodb-mcp-server - MongoDB's official MCP server. The tool implementations (find, aggregate, count, explain, collection-schema) are adapted from this project with security hardening for public deployment.