Lightweight Self-Hosted AI Web Search MCP Solution

Lightweight Self-Hosted AI Web Search MCP Solution

This repository provides a lightweight, self-hosted MCP solution designed for AI developers requiring efficient web search capabilities. It integrates with multiple search engines and offers local caching and LLM-based summarization, streamlining the process of gathering and processing information for AI applications. This tool is ideal for developers who need a controlled, private, and performant web search component within their AI workflows.

What it Does

The websearch-mcpserver acts as a centralized agent that queries various web search engines on behalf of your AI. It supports popular engines like Baidu, Tavily, and Bing , as well as academic search platforms. Results are processed locally, with an integrated SQLite cache to store previous queries and results, reducing redundant API calls and improving response times. Furthermore, it leverages Large Language Models (LLMs) to provide concise summaries of search results , making information more digestible for AI agents.

Key Features

• Multi-Engine Support: Integrates with Baidu, Tavily, Bing, and academic search engines.
• Self-Hosted: Full control over data and infrastructure.
• SQLite Caching: Efficient local storage of search queries and results.
• LLM Summarization: Generates concise summaries of web content.
• Lightweight Architecture: Designed for minimal resource consumption.
• MCP Protocol: Compatible with existing MCP server setups.

Who it's For

This tool is specifically built for AI developers who are:

• Building AI agents that require real-time web information.
• Prioritizing data privacy and control through self-hosting.
• Seeking to optimize search performance and reduce API costs.
• Working with existing MCP server architectures.
• Developing applications that benefit from summarized web content.