one-search-mcp

one-search-mcp

24
browser_automation

OneSearch MCP Server is an MCP implementation providing integrated web search, crawling, and data extraction functionalities. It supports multiple search engines and is suitable for local and self-hosted environments.

🚀 OneSearch MCP Server: Web Search & Crawl & Scraper & Extract

A Model Context Protocol (MCP) server implementation that integrates with search engines such as Searxng, Tavily, DuckDuckGo, and Bing for web search and scraping with Firecrawl.

Features

  • Perform web searches, scraping, crawling, and content extraction from various websites.
  • Supports multiple search engines and web scrapers including SearXNG, Firecrawl, Tavily, DuckDuckGo, and Bing.
  • Enable local web searches with browser support for Bing, Google, Baidu, and Sogou.
  • Use "puppeteer-core" for website content scraping.
  • Tools provided include one_search, one_scrape, and one_map.
  • Self-hosting support for SearXNG and Firecrawl.

Installation

  • Install via Smithery or manually with npm.

Environment Variables

  • Configure search provider, API URLs, and keys for SearxNG and Firecrawl connections.

Running on Cursor and Windsurf

  • Instructions for configuring mcp.json and model_config.json for specific environments.

Self-host

  • Instructions available in the deploy README.

Troubleshooting

  • Addresses a Puppeteer issue when using tsx.