WebSearch-MCP is a self-hosted Model Context Protocol (MCP) server that provides web search capabilities to AI assistants supporting MCP. It enables AI models such as Claude to search the web in real time and obtain the latest information on any topic. The server integrates with a Crawler API service, which performs the actual web search, and communicates with AI assistants using the standardized MCP protocol.
Purpose: To provide AI assistants with real-time web search functionality so they can answer questions that require the latest information.
Main capabilities: Exposes a web_search tool through the stdio transport, supporting parameters such as query, number of results, language, region, domain include/exclude, term exclusion, and result type (all, news, blogs).
Usage: The crawler service must first be set up via Docker Compose, and then the MCP client (such as Claude Desktop, Cursor IDE, Cline) must be configured. It can be installed automatically via Smithery, or installed manually, or used directly without installation. It supports configuring API_URL and MAX_SEARCH_RESULT via environment variables.
Applicable scenarios: Integration with AI assistants that require real-time search capabilities.
Limitations: The crawler service must be self-hosted; some websites have Cloudflare protection and may require FlareSolverr; specific fees, pricing, effectiveness, etc. are not yet provided.
