🤖 Google SERP & AI Agent Grounding — Model Context Protocol (MCP) Server

A high-performance Model Context Protocol (MCP) Server and live Google search grounding engine for Claude Desktop, Cursor IDE, LangChain, and Autonomous AI Agents.

⚡ Overview

This repository provides a standardized Model Context Protocol (MCP) server (mcp_server.py) that equips LLMs and autonomous agents with real-time web search and webpage content extraction without hallucination.

• 🛡️ Zero 429 Rate-Limit Blocks: Automatic anti-blocking engine with multi-engine fallback.

• 🤖 LLM-Optimized Grounding: Returns token-budgeted Markdown context with structured citations ([1], [2]).

• 📄 Universal Markdown Extractor: Strips navigation, ads, and scripts from any URL to produce dense Markdown for RAG context windows.

• 🔌 Zero-Config MCP Integration: Plug-and-play with Claude Desktop, Cursor, and Antigravity.

Related MCP server: Advanced Web Search MCP Server

🛠️ MCP Tools & Capabilities

The server exposes the following Model Context Protocol (MCP) tools:

🔌 Quickstart: Connect to Claude Desktop & Cursor

1. Claude Desktop Configuration

Add this server to your claude_desktop_config.json:

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json

• Windows: %APPDATA%\Claude\claude_desktop_config.json

• Linux: ~/.config/Claude/claude_desktop_config.json

{
  "mcpServers": {
    "google-serp-agent": {
      "command": "python3",
      "args": [
        "/path/to/google-serp-extractor/mcp_server.py"
      ]
    }
  }
}

2. Cursor IDE Configuration (.cursor/mcp.json)

{
  "mcpServers": {
    "google-serp-agent": {
      "command": "python3",
      "args": ["/path/to/google-serp-extractor/mcp_server.py"]
    }
  }
}

📦 Python SDK & LangChain Tool Usage

You can also use this library directly inside Python scripts and LangChain agents:

pip install google-serp-extractor

from google_serp_extractor import agent_search_and_ground, extract_webpage_markdown

# 1. Ground your LLM with real-time web search facts
result = agent_search_and_ground(
    query="Latest breakthroughs in Quantum Computing 2026",
    max_results=3,
    max_tokens=1500
)

# Feed context directly into your LLM prompt
print(result["context_markdown"])

# 2. Extract clean Markdown from any live URL
page_md = extract_webpage_markdown("https://en.wikipedia.org/wiki/Artificial_intelligence")
print(page_md["markdown"][:500])

🌐 Serverless REST API Endpoints

📄 License

MIT License. Created by Meanus Arcanus.