Socket MCP Server

A Model Context Protocol (MCP) server for Socket integration, allowing AI assistants to efficiently check dependency vulnerability scores and security information.

✨ Features

• 🔍 Dependency Security Scanning - Get comprehensive security scores for npm, PyPI, and other package ecosystems

• 🌐 Public Hosted Service - Use our public server at https://mcp.socket.dev/ with no setup required

• 🚀 Multiple Deployment Options - Run locally via stdio, HTTP, or use our service

• 🤖 AI Assistant Integration - Works seamlessly with Claude, VS Code Copilot, Cursor, and other MCP clients

• 📊 Batch Processing - Check multiple dependencies in a single request

• 🔒 No Authentication Required - Public server requires no API keys or registration

🛠️ This project is in early development and rapidly evolving.

Related MCP server: MCP NVD Server

🚀 Quick Start

Option 1: Use the Public Socket MCP Server (Recommended)

The easiest way to get started is to use our public Socket MCP server. No API key or authentication required! Click a button below to install the public server in your favorite AI assistant.

To use the public Socket MCP server with Claude Desktop:

1. In Claude Desktop, go to Settings > Developer > Edit Config.

2. Add the Socket MCP server configuration:

{
  "mcpServers": {
    "socket-mcp": {
      "type": "http",
      "url": "https://mcp.socket.dev/"
    }
  }
}

3. Save the configuration and restart Claude Desktop.

4. Now you can ask Claude questions like "Check the security score for express version 4.18.2".

The process is similar for Claude Code. See the Claude Code documentation for more details. Here's an example command to add the Socket MCP server:

claude mcp add --transport http socket-mcp https://mcp.socket.dev/

You can install the Socket MCP server using the VS Code CLI:

# For VS Code with GitHub Copilot
code --add-mcp '{"name":"socket-mcp","type":"http","url":"https://mcp.socket.dev/}'

After installation, the Socket MCP server will be available for use with your GitHub Copilot agent in VS Code.

Alternatively, you can manually add it to your VS Code MCP configuration in .vscode/mcp.json:

{
  "servers": {
    "socket-mcp": {
      "type": "http",
      "url": "https://mcp.socket.dev/"
    }
  }
}

Go to Cursor Settings -> MCP -> Add new MCP Server. Name it "socket-mcp", use http type with URL https://mcp.socket.dev/.

{
  "mcpServers": {
    "socket-mcp": {
      "type": "http",
      "url": "https://mcp.socket.dev/"
    }
  }
}

To use the Socket MCP server in Windsurf:

1. Open Windsurf Settings

2. Navigate to MCP Servers section

3. Add a new server with the following configuration:

{
    "mcpServers": {
        "socket-mcp": {
            "serverUrl": "https://mcp.socket.dev/mcp"
        }
    }
}

4. Save the configuration and restart Windsurf if needed.

Option 2: Deploy Socket MCP Server on your machine

If you prefer to run your own instance, you can deploy the Socket MCP server locally using either stdio or HTTP modes.

Getting an API key

To use a local Socket MCP Server, you need to create an API key. You can do this by following these steps. The only required permission scope is packages:list, which allows the MCP server to query package metadata for dependency scores.

For local deployment, you have two options:

Option 2a: Stdio Mode (Default)

Click a button below to install the self-hosted stdio server in your favorite AI assistant.

Claude Code (stdio mode) can be set up with the following command:

claude mcp add socket-mcp -e SOCKET_API_KEY="your-api-key-here" -- npx -y @socketsecurity/mcp@latest

This is how the configuration looks like on most MCP clients:

{
  "mcpServers": {
    "socket-mcp": {
      "command": "npx",
      "args": ["@socketsecurity/mcp@latest"],
      "env": {
        "SOCKET_API_KEY": "your-api-key-here"
      }
    }
  }
}

This approach automatically uses the latest version without requiring global installation.

Option 2b: HTTP Mode

1. Run the server in HTTP mode using npx:

   MCP_HTTP_MODE=true SOCKET_API_KEY=your-api-key npx @socketsecurity/mcp@latest --http

   HTTP mode supports these environment variables:

   Variable	Required	Default	Description
   SOCKET_API_KEY	Required unless OAuth is enabled	None	Socket API key used for outbound API calls. If unset in OAuth-enabled HTTP mode, the validated incoming bearer token is forwarded upstream instead.
   SOCKET_OAUTH_ISSUER	Set together with the two introspection vars to enable OAuth	None	OAuth issuer URL used for metadata discovery and incoming bearer-token validation.
   SOCKET_OAUTH_INTROSPECTION_CLIENT_ID	With OAuth	None	Client ID used for token introspection.
   SOCKET_OAUTH_INTROSPECTION_CLIENT_SECRET	With OAuth	None	Client secret used for token introspection.
   SOCKET_OAUTH_REQUIRED_SCOPES	No	packages:list	Space-delimited scopes required on incoming access tokens.
   SOCKET_API_URL	No	Production Socket API URL, or localhost when SOCKET_DEBUG=true	Override the upstream Socket API endpoint. Useful for local development and testing.
   SOCKET_DEBUG	No	false	Switches the default upstream Socket API endpoint to localhost when SOCKET_API_URL is unset.
   TRUST_PROXY	No	false	When true, trust X-Forwarded-Host and X-Forwarded-Proto when building OAuth metadata URLs. Enable only behind a trusted reverse proxy that rewrites these headers.
   MCP_PORT	HTTP mode only	3000	Port to bind the HTTP server to.

   SOCKET_API_URL and SOCKET_DEBUG also apply in stdio mode. In OAuth-enabled HTTP mode, if SOCKET_API_KEY is unset, the authenticated client's bearer token is forwarded to the Socket API. That token therefore must also be accepted by the configured upstream Socket API.

   To enable OAuth-backed auth for incoming MCP requests:

   MCP_HTTP_MODE=true \
   SOCKET_OAUTH_ISSUER=https://issuer.example.com \
   SOCKET_OAUTH_INTROSPECTION_CLIENT_ID=your-client-id \
   SOCKET_OAUTH_INTROSPECTION_CLIENT_SECRET=your-client-secret \
   npx @socketsecurity/mcp@latest --http

   Add TRUST_PROXY=true only when the server is deployed behind a trusted reverse proxy or load balancer that normalizes the forwarded host and protocol headers.

2. Configure your MCP client to connect to the HTTP server:

   {
     "mcpServers": {
       "socket-mcp": {
         "type": "http",
         "url": "http://localhost:3000"
       }
     }
   }

Tools exposed by the Socket MCP Server

depscore

The depscore tool allows AI assistants to query the Socket API for dependency scoring information. It provides comprehensive security and quality metrics for packages across different ecosystems.

Parameters:

Example Usage:

{
  "packages": [
    {
      "ecosystem": "npm",
      "depname": "express",
      "version": "4.18.2"
    },
    {
      "ecosystem": "pypi",
      "depname": "fastapi",
      "version": "0.100.0"
    }
  ]
}

Sample Response:

pkg:npm/express@4.18.2: supply_chain: 1.0, quality: 0.9, maintenance: 1.0, vulnerability: 1.0, license: 1.0
pkg:pypi/fastapi@0.100.0: supply_chain: 1.0, quality: 0.95, maintenance: 0.98, vulnerability: 1.0, license: 1.0

How to Use the Socket MCP Server

1. Ask your AI assistant to check dependencies:

   • "Check the security score for express version 4.18.2"

   • "Analyze the security of my package.json dependencies"

   • "What are the vulnerability scores for react, lodash, and axios?"

2. Get comprehensive security insights including supply chain, quality, maintenance, vulnerability, and license scores.

Adjust tool usage with custom rules

You can further customize how the Socket MCP server interacts with your AI assistant by modifying your client rules. The rules are usually a markdown file and its location depends on the AI assistant you are using.

Rules that can be added to the client rules file include the following:

Always check dependency scores with the depscore tool when you add a new dependency. If the score is low, consider using an alternative library or writing the code yourself. If you are unsure about the score, ask for a review from someone with more experience. When checking dependencies, make sure to also check the imports not just the pyproject.toml/package.json/dependency file.

You can adjust the rules to fit your needs. For example, you can add rules to include specific manifest files, or guide the AI assistant on how to handle low scores. The rules are flexible and can be tailored to your workflow.

Development

For End Users

For most users, we recommend using either:

1. Public server: https://mcp.socket.dev/ (no setup required)

2. NPX command: npx @socketsecurity/mcp@latest (always latest version)

For Contributors

If you want to contribute to the Socket MCP server development:

Health Check Endpoint

When running in HTTP mode, the server provides a health check endpoint for Kubernetes and Docker deployments:

GET /health

Response:

{
  "status": "healthy",
  "service": "socket-mcp",
  "version": "0.0.3",
  "timestamp": "2025-06-17T20:45:22.059Z"
}

This endpoint can be used for:

• Kubernetes liveness and readiness probes

• Docker health checks

• Load balancer health monitoring

• General service monitoring

Prerequisites

• Node.js v16 or higher

• npm or yarn

Installation

Clone the repository and install dependencies:

git clone https://github.com/SocketDev/socket-mcp.git
cd socket-mcp
npm install

Build

This project is a directly runnable Node.js project using Type stripping. If you are on Node.js 22, run with node --experimental-strip-types index.ts. On any later versions of Node.js, you can simply run node index.ts. In either version you can also run the npm run scripts which include the correct flags.

The js files will automatically be build when running npm publish, and cleaned up afterwards with npm run clean.

If you want to preview the build you can run:

npm run build

Run from Source

To run the Socket MCP server from source:

export SOCKET_API_KEY=your_api_key_here
node --experimental-strip-types index.ts

Or in HTTP mode:

MCP_HTTP_MODE=true SOCKET_API_KEY=your_api_key_here node --experimental-strip-types index.ts --http

🔧 Troubleshooting

Common Issues

Q: The public server isn't responding

• Check that you're using the correct URL: https://mcp.socket.dev/

• Verify your MCP client configuration is correct

• Try restarting your MCP client

Q: Local server fails to start

• Ensure you have Node.js v16+ installed

• Check that your SOCKET_API_KEY environment variable is set

• Verify the API key has packages:list permission

Q: Getting authentication errors with local server

• Double-check your Socket API key is valid

• Ensure the key has the required packages:list scope

• Try regenerating your API key from the Socket dashboard

Q: AI assistant can't find the depscore tool

• Restart your MCP client after configuration changes

• Verify the server configuration is saved correctly

• Check that the MCP server is running (for local deployments)

Getting Help

• 📖 Socket Documentation

• 🐛 Report Issues

• 💬 Community Support