SearXNG

关于

searXNG-mcp is a FastMCP-based server designed to provide powerful web search and scraping capabilities, specifically optimized for integration with Large Language Models (LLMs). It leverages a running SearXNG instance for multi-query web searches and offers dual scraping methods—fast HTTP requests for static content and full browser rendering with Playwright for dynamic JavaScript-heavy pages. A key innovation is its use of TOON encoding/decoding, which ensures token-efficient communication, making the retrieved data more suitable for LLM processing. The server boasts a modular architecture, robust retry logic, comprehensive logging, and configurable environment variables, making it a reliable tool for augmenting AI agents with real-time web data.

主要功能

• 0 GitHub stars
• Multi-query web search via SearXNG with configurable result counts
• Dual scraping methods (requests for static, browser for dynamic JavaScript pages)
• TOON encoding for token-efficient LLM communication
• Persistent browser instance reuse for better performance
• Retry logic with exponential backoff for robust operation

使用案例

• Integrating advanced web data retrieval into applications requiring efficient LLM communication
• Collecting content from dynamic, JavaScript-rendered websites for data analysis or AI training
• Empowering LLMs and AI agents with real-time, up-to-date web search results