LinkedIn Network MCP

Local npm MCP server that fetches LinkedIn network data from your backend server and exports to CSV.

Architecture

Cursor (MCP Client)
    ↓ npx
Local MCP Server (server.py)
    ↓ HTTP GET /api/network (with X-UUID header)
Backend Server (Railway)
    ↓ Queries database
PostgreSQL Database
    ↓ Returns data
Backend Server
    ↓ Returns JSON response
Local MCP Server
    ↓ Converts to CSV
CSV File (saved locally)

Important: The MCP server does NOT connect directly to the database. It only communicates with your backend server via HTTP API.

Quick Start

1. Make sure Python 3.8+ is installed (dependencies will be installed automatically)

2. Configure ~/.cursor/mcp.json:

   {
     "mcpServers": {
       "network-mcp-node": {
         "command": "npx",
         "args": ["-y", "@ghostteam/network-mcp-node@latest"],
         "env": {
           "UUID": "your-uuid-here",
           "OUT_DIR": "C:\\Users\\User\\Documents"
         }
       }
     }
   }

3. Restart Cursor

4. Use the export_network_csv_to_file MCP tool

How It Works

1. MCP Client (Cursor) runs npx @ghostteam/network-mcp-node@latest
2. Local MCP Server starts (Python)
3. User calls export tool → Server sends HTTP GET to backend:

   GET /api/network
   Headers: X-UUID: your-uuid-here

4. Backend Server queries database for that UUID
5. Backend returns JSON array of contacts
6. MCP Server saves CSV file to OUT_DIR

Backend API Required

Your backend server needs this endpoint:

GET /api/network
Headers:
  X-UUID: <user-uuid>

Response:
  [
    {
      "full_name": "...",
      "email": "...",
      ...
    },
    ...
  ]

Environment Variables

• UUID (required): User identifier sent to backend server
• OUT_DIR (optional): Where to save CSV files (defaults to current directory)

Note: Backend URL is hardcoded in the package and cannot be changed via configuration.

Publishing to npm

npm login
npm publish --access public