Mcp Mongo Server

What is Mcp Mongo Server about?

Mcp Mongo Server provides a Model Context Protocol (MCP) endpoint that lets large language models query, aggregate, and modify MongoDB collections through a uniform JSON‑based API. It abstracts connection details, automatically converts string IDs to ObjectId when appropriate, and can enforce read‑only operation for production safety.

How to use Mcp Mongo Server?

1. Installation – install globally with npm install -g mcp-mongo-server or run directly via npx.
2. Start the server – supply a MongoDB URI either as a command‑line argument or through the MCP_MONGODB_URI environment variable. Add --read-only (or set MCP_MONGODB_READONLY=true) to disable write operations.

   npx -y mcp-mongo-server mongodb://user:pass@localhost:27017/mydb --read-only
   

3. Configure in an LLM client – add an entry to the client’s mcpServers configuration (Claude Desktop, Windsurf, Cursor, etc.) pointing to the npx command.
4. Invoke tools – the client sends JSON tool calls such as query, aggregate, update, insert, or createIndex. The server returns results or errors over stdio.

Key Features

• Smart ObjectId handling – auto, none, or force modes for converting string IDs.
• Flexible configuration – environment variables and CLI flags for URI and read‑only mode.
• Read‑only mode – protects production databases by routing reads to secondaries and blocking writes.
• Full CRUD support – query, aggregation, count, schema introspection, update, insert, and index creation.
• LLM‑centric tools – collection completions, schema inference, and analysis to improve model context.
• Docker ready – official Dockerfile and example docker run commands.
• Debugging support – optional MCP Inspector for visual debugging.

Use Cases

• Chat‑based database assistants – let LLMs answer user questions about data without exposing raw credentials.
• Automated data analysis – run aggregation pipelines triggered by natural‑language prompts.
• Safe production access – expose a read‑only endpoint for monitoring dashboards powered by LLMs.
• Rapid prototyping – developers can mock MongoDB interactions in Claude Desktop or similar tools.
• CI/CD validation – run integration tests that call the server’s tools to verify schema and query correctness.

FAQ

Q: Do I need to rebuild after changing code? A: No. When using the published npm package or the Docker image, changes are already compiled. For local development use npm run watch to auto‑rebuild.

Q: How does read‑only mode work? A: The server sets the MongoDB read preference to secondaryPreferred and disables any write‑related tool calls (update, insert, createIndex).

Q: Can I force all ID fields to ObjectId? A: Yes, set objectIdMode to "force" via the server’s configuration (environment variable or future CLI flag).

Q: What if my collection uses custom primary keys? A: Use objectIdMode: "none" to leave string IDs untouched.

Q: Is the server compatible with cloud MongoDB services? A: Absolutely – any standard MongoDB URI (Atlas, mLab, self‑hosted) works.

MCP MongoDB Server

---

A Model Context Protocol server that enables LLMs to interact with MongoDB databases. This server provides capabilities for inspecting collection schemas and executing MongoDB operations through a standardized interface.

Demo

Key Features

Smart ObjectId Handling

• Intelligent conversion between string IDs and MongoDB ObjectId
• Configurable with objectIdMode parameter:
  • "auto": Convert based on field names (default)
  • "none": No conversion
  • "force": Force all string ID fields to ObjectId

Flexible Configuration

• Environment Variables:
  • MCP_MONGODB_URI: MongoDB connection URI
  • MCP_MONGODB_READONLY: Enable read-only mode when set to "true"
• Command-line Options:
  • --read-only or -r: Connect in read-only mode

Read-Only Mode

• Protection against write operations (update, insert, createIndex)
• Uses MongoDB's secondary read preference for optimal performance
• Ideal for safely connecting to production databases

MongoDB Operations

• Read Operations:
  • Query documents with optional execution plan analysis
  • Execute aggregation pipelines
  • Count documents matching criteria
  • Get collection schema information
• Write Operations (when not in read-only mode):
  • Update documents
  • Insert new documents
  • Create indexes

LLM Integration

• Collection completions for enhanced LLM interaction
• Schema inference for improved context understanding
• Collection analysis for data insights

Installation

Global Installation

npm install -g mcp-mongo-server

For Development

# Clone repository
git clone https://github.com/kiliczsh/mcp-mongo-server.git
cd mcp-mongo-server

# Install dependencies
npm install

# Build
npm run build

# Development with auto-rebuild
npm run watch

Usage

Basic Usage

# Start server with MongoDB URI
npx -y mcp-mongo-server mongodb://muhammed:kilic@localhost:27017/database

# Connect in read-only mode
npx -y mcp-mongo-server mongodb://muhammed:kilic@localhost:27017/database --read-only

Environment Variables

You can configure the server using environment variables, which is particularly useful for CI/CD pipelines, Docker containers, or when you don't want to expose connection details in command arguments:

# Set MongoDB connection URI
export MCP_MONGODB_URI="mongodb://muhammed:kilic@localhost:27017/database"

# Enable read-only mode
export MCP_MONGODB_READONLY="true"

# Run server (will use environment variables if no URI is provided)
npx -y mcp-mongo-server

Using environment variables in Claude Desktop configuration:

{
  "mcpServers": {
    "mongodb-env": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-mongo-server"
      ],
      "env": {
        "MCP_MONGODB_URI": "mongodb://muhammed:kilic@localhost:27017/database",
        "MCP_MONGODB_READONLY": "true"
      }
    }
  }
}

Using environment variables with Docker:

# Build
docker build -t mcp-mongo-server .

# Run
docker run -it -d -e MCP_MONGODB_URI="mongodb://muhammed:kilic@localhost:27017/database" -e MCP_MONGODB_READONLY="true" mcp-mongo-server

# or edit docker-compose.yml and run
docker-compose up -d

Integration with Claude Desktop

Manual Configuration

Add the server configuration to Claude Desktop's config file:

MacOS: ~/Library/Application Support/Claude/claude_desktop_config.json Windows: %APPDATA%/Claude/claude_desktop_config.json

Command-line Arguments Approach:

{
  "mcpServers": {
    "mongodb": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-mongo-server",
        "mongodb://muhammed:kilic@localhost:27017/database"
      ]
    },
    "mongodb-readonly": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-mongo-server",
        "mongodb://muhammed:kilic@localhost:27017/database",
        "--read-only"
      ]
    }
  }
}

Environment Variables Approach:

{
  "mcpServers": {
    "mongodb": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-mongo-server"
      ],
      "env": {
        "MCP_MONGODB_URI": "mongodb://muhammed:kilic@localhost:27017/database"
      }
    },
    "mongodb-readonly": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-mongo-server"
      ],
      "env": {
        "MCP_MONGODB_URI": "mongodb://muhammed:kilic@localhost:27017/database",
        "MCP_MONGODB_READONLY": "true"
      }
    }
  }
}

GitHub Package Usage:

{
  "mcpServers": {
    "mongodb": {
      "command": "npx",
      "args": [
        "-y",
        "github:kiliczsh/mcp-mongo-server",
        "mongodb://muhammed:kilic@localhost:27017/database"
      ]
    },
    "mongodb-readonly": {
      "command": "npx",
      "args": [
        "-y",
        "github:kiliczsh/mcp-mongo-server",
        "mongodb://muhammed:kilic@localhost:27017/database",
        "--read-only"
      ]
    }
  }
}

Integration with Windsurf and Cursor

The MCP MongoDB Server can be used with Windsurf and Cursor in a similar way to Claude Desktop.

Windsurf Configuration

Add the server to your Windsurf configuration:

{
  "mcpServers": {
    "mongodb": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-mongo-server",
        "mongodb://muhammed:kilic@localhost:27017/database"
      ]
    }
  }
}

Cursor Configuration

For Cursor, add the server configuration to your settings:

{
  "mcpServers": {
    "mongodb": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-mongo-server",
        "mongodb://muhammed:kilic@localhost:27017/database"
      ]
    }
  }
}

You can also use the environment variables approach with both Windsurf and Cursor, following the same pattern shown in the Claude Desktop configuration.

Automated Installation

Using Smithery:

npx -y @smithery/cli install mcp-mongo-server --client claude

Using mcp-get:

npx @michaellatman/mcp-get@latest install mcp-mongo-server

Available Tools

Query Operations

• query: Execute MongoDB queries

  {
    collection: "users",
    filter: { age: { $gt: 30 } },
    projection: { name: 1, email: 1 },
    limit: 20,
    explain: "executionStats"  // Optional
  }
  

• aggregate: Run aggregation pipelines

  {
    collection: "orders",
    pipeline: [
      { $match: { status: "completed" } },
      { $group: { _id: "$customerId", total: { $sum: "$amount" } } }
    ],
    explain: "queryPlanner"  // Optional
  }
  

• count: Count matching documents

  {
    collection: "products",
    query: { category: "electronics" }
  }
  

Write Operations

• update: Modify documents

  {
    collection: "posts",
    filter: { _id: "60d21b4667d0d8992e610c85" },
    update: { $set: { title: "Updated Title" } },
    upsert: false,
    multi: false
  }
  

• insert: Add new documents

  {
    collection: "comments",
    documents: [
      { author: "user123", text: "Great post!" },
      { author: "user456", text: "Thanks for sharing" }
    ]
  }
  

• createIndex: Create collection indexes

  {
    collection: "users",
    indexes: [
      {
        key: { email: 1 },
        unique: true,
        name: "email_unique_idx"
      }
    ]
  }
  

System Operations

• serverInfo: Get MongoDB server details

  {
    includeDebugInfo: true  // Optional
  }
  

Debugging

Since MCP servers communicate over stdio, debugging can be challenging. Use the MCP Inspector for better visibility:

npm run inspector

This will provide a URL to access the debugging tools in your browser.

Running evals

The evals package loads an mcp client that then runs the index.ts file, so there is no need to rebuild between tests. You can load environment variables by prefixing the npx command. Full documentation can be found here.

OPENAI_API_KEY=your-key  npx mcp-eval src/evals/evals.ts src/schemas/tools.ts

License

This MCP server is licensed under the MIT License. This means you are free to use, modify, and distribute the software, subject to the terms and conditions of the MIT License. For more details, please see the LICENSE file in the project repository.