Open-WebSearch MCP

Example Use Cases

• 1

  Perform up-to-date web searches across multiple search engines without requiring API keys or external credentials.

• 2

  Fetch and extract full article or README content from sites like CSDN, Juejin, and GitHub for downstream processing or summarization.

• 3

  A locally hostable, configurable MCP endpoint (HTTP/SSE/STDIO) with proxy and CORS support for integration into local workflows or private deployments.

Description

Open-WebSearch MCP server scrapes and aggregates search results from multiple engines (bing, duckduckgo, exa, brave, baidu, csdn, juejin) and returns structured results (title, url, description). It supports fetching full article content for certain sites (CSDN, Juejin, GitHub READMEs) and can be run locally, via npx, or in Docker with optional HTTP proxy and CORS configuration. The server is designed for integration with MCP-capable agents and tools (HTTP/SSE/STDIO transports) to enable live web lookups without third-party API keys. It emphasizes configurability (default engine, allowed engines, result limits) and ease of deployment.