wenmin-wu/google-search-ai-mcp

Google Search AI MCP

A MCP server and Chrome extension that enables AI agents to conduct conversations with Google Search AI through browser automation.

📸 Demo

🌟 Features

✅ Simple Setup: WebSocket-based communication ✅ FastMCP Integration: Uses stdio transport for seamless MCP client integration
✅ Chrome Extension: Automated Google Search AI interaction
✅ HTML Parsing: Converts Google Search AI responses to clean markdown
✅ Multi-Round Conversations: Supports ongoing conversations with context

🚀 Quick Start

1. Install Python Dependencies

cd google-search-ai-mcp
pip install -r requirements.txt

2. Start the MCP Server

# Start the server with default WebSocket port 8761
python server.py

# Or specify custom WebSocket port
WEBSOCKET_PORT=9000 python server.py

You should see output like:

🚀 Starting Google Search AI MCP Server (Community Version)
🔌 WebSocket Server: ws://0.0.0.0:8761
📡 FastMCP: stdio transport
✅ WebSocket server running on ws://0.0.0.0:8761

3. Install Chrome Extension

1. Open Chrome and go to chrome://extensions/
2. Enable "Developer mode" (toggle in top right)
3. Click "Load unpacked"
4. Select the google-search-ai-mcp/chrome-extension/ folder
5. The extension should appear with a Google AI icon

4. Configure the Extension

1. Click the extension icon in Chrome toolbar
2. Click "Configure" if you need to change the WebSocket URL
3. Click "Open Google AI" to open a Google Search AI tab
4. The extension will automatically connect to the MCP server

🧪 Usage with AI Agents

MCP Client Configuration

Add to your MCP client configuration:

{
  "mcpServers": {
    "google-search-ai": {
      "command": "$(which python)",
      "args": ["/path/to/google-search-ai-mcp/server.py"]
    }
  }
}

💡 Tip: Use the full Python path (e.g., /opt/anaconda3/envs/py310/bin/python3) instead of python to ensure the correct Python environment with installed packages is used. Run which python to find your Python path.

Available MCP Tool

The system provides a single, powerful tool:

chat_search_ai

Chat with Google Search AI with automatic conversation management.

Parameters:

• message: Message to send to Google Search AI. Say 'done' to end the conversation.

📁 Project Structure

google-search-ai-mcp/
├── server.py              # FastMCP server with WebSocket support
├── requirements.txt       # Python dependencies
├── README.md              # This file
└── chrome-extension/      # Chrome extension files
    ├── manifest.json      # Extension manifest
    ├── background.js      # WebSocket client & tab management
    ├── content.js         # Google Search AI automation
    ├── popup.html         # Extension popup UI
    ├── popup.js           # Popup logic
    └── icon*.png          # Extension icons

🔧 Configuration

WebSocket Server

The server accepts these environment variables:

• WEBSOCKET_PORT: WebSocket server port (default: 8761)

Chrome Extension

Configure the WebSocket URL in the extension popup:

1. Click extension icon → "Configure"
2. Enter your WebSocket URL (default: ws://localhost:8761)
3. Click "Save"

🐛 Troubleshooting

Common Issues

Extension Won't Connect

• Problem: Popup shows "Disconnected"
• Solution:
  • Ensure MCP server is running on port 8761
  • Check Chrome console for WebSocket errors
  • Verify firewall isn't blocking the connection

Google Search AI Not Found

• Problem: Content script errors about missing elements
• Solution:
  • Navigate to https://www.google.com/search?udm=50 manually
  • Ensure you're logged into Google
  • Check that Google Search AI is available in your region

MCP Server Not Starting

• Problem: Import errors or dependency issues
• Solution:

  # Reinstall dependencies
  pip install --force-reinstall -r requirements.txt
  
  # Check Python version (3.8+ required)
  python --version
  

Wrong Python Path in MCP Configuration

• Problem: MCP client uses system Python instead of environment with installed packages

• Solution: Use the full Python path in your MCP configuration:

  # Find your Python path
  which python
  # Or for Python 3 specifically
  which python3
  

  Then update your MCP configuration to use the full path:

  {
    "mcpServers": {
      "google-search-ai": {
        "command": "/opt/anaconda3/envs/py310/bin/python3",
        "args": ["/path/to/google-search-ai-mcp/server.py"]
      }
    }
  }
  

  Common Python paths:

  • Conda: /opt/anaconda3/envs/your-env/bin/python3
  • Homebrew: /opt/homebrew/bin/python3
  • System: /usr/bin/python3
  • Virtual env: /path/to/venv/bin/python3

Debug Mode

Enable debug logging in the server:

# Set debug level
LOG_LEVEL=DEBUG python server.py

Testing WebSocket Connection

# Install wscat if you don't have it
npm install -g wscat

# Connect to WebSocket server
wscat -c ws://localhost:8761

# Send test message
{"type": "connection_test", "data": {"ping": "test"}}