google-ai-scraper

MCP server for scraping Google AI Overviews via a Chrome extension relay. No API keys needed. It uses your own browser session and a shared local FastAPI backend.

How it works

MCP Client (Claude Code, Cursor, etc.)
    ↕ stdio or SSE
MCP Server (this package)
    ↕ HTTP
Shared FastAPI Backend (:15551)
    ↕ polling
Chrome Extension
    → opens background tab on google.com
    → scrapes AI Overview → Markdown
    → posts result back

In default stdio mode, each MCP client process auto-reuses or auto-starts the shared backend on 127.0.0.1:15551.

Install

# Via uvx (recommended)
uvx google-ai-scraper

# Or pip
pip install google-ai-scraper

Prerequisites

1. Chrome with the Google AI Overview Scraper extension installed
2. Python 3.10+
3. Extension options pointed at http://localhost:15551

MCP Client Configuration

Claude Code

{
  "mcpServers": {
    "google-ai-scraper": {
      "command": "uvx",
      "args": ["google-ai-scraper"]
    }
  }
}

Claude Desktop

{
  "mcpServers": {
    "google-ai-scraper": {
      "command": "uvx",
      "args": ["google-ai-scraper"]
    }
  }
}

Cursor

{
  "mcpServers": {
    "google-ai-scraper": {
      "command": "uvx",
      "args": ["google-ai-scraper"]
    }
  }
}

Local Source Checkout

{
  "mcpServers": {
    "google-ai-scraper": {
      "command": "uv",
      "args": [
        "run",
        "--directory",
        "/absolute/path/to/server",
        "google-ai-scraper"
      ]
    }
  }
}

MCP Tools

CLI Options

google-ai-scraper [--sse] [--no-server] [--backend] [--port PORT]

Environment Variables

Advanced Modes

Explicit Shared Backend

uv run google-ai-scraper --backend --port 15551

SSE Server

uv run google-ai-scraper --sse --no-server

Shared Backend + SSE Helper Script

cd server
./start-services.sh

This starts the shared backend on :15551 and the SSE endpoint on :8001.

License

MIT