WebSearch MCP
Real-time web search for MCP clients via a self-hosted crawler API over stdio.
Add to your client
Copy the config for your MCP client and paste it into its config file.
npm install -g websearch-mcpPaste into ~/Library/Application Support/Claude/claude_desktop_config.json
{
"mcpServers": {
"websearch-mcp": {
"command": "npx",
"args": [
"websearch-mcp"
],
"env": {
"API_URL": "http://localhost:3001",
"MAX_SEARCH_RESULT": "5"
}
}
}
}Step-by-step guides: Add to Claude Desktop · Add to Cursor · Add to Windsurf
Before you start
- Node.js with npx (to run the websearch-mcp package)
- A running WebSearch Crawler API backend (Docker image laituanmanh/websearch-crawler) reachable at API_URL
- FlareSolverr service (for bypassing Cloudflare protection), started via the provided docker-compose.yml
- Docker and Docker Compose to run the crawler service
About WebSearch MCP
An MCP server that provides web search to AI assistants like Claude over stdio. It forwards queries to a self-hosted WebSearch Crawler API (Docker + FlareSolverr) and returns structured results (title, snippet, url, siteName, byline). Configured via the API_URL and MAX_SEARCH_RESULT environment variables.
Tools & capabilities (1)
web_searchSearch the web in real time. Parameters: query (required); numResults (optional, default 5); language (optional, e.g. 'en'); region (optional, e.g. 'us'); excludeDomains (optional); includeDomains (optional); excludeTerms (optional); resultType (optional: 'all', 'news', or 'blogs'). Returns results with title, snippet, url, siteName, and byline.
What this server can do
WebSearch MCP provides tools for these capabilities — tap one to see every MCP server that does the same:
When to use it
- Give Claude or other MCP-aware assistants up-to-date information from the live web
- Run a privacy-friendly, self-hosted web search backend instead of a paid search API
- Filter search results by domain, term, language/region, or content type (news/blogs)
- Bypass Cloudflare-protected sites during crawling via FlareSolverr
Security notes
Requires a self-hosted WebSearch Crawler API backend (Docker image laituanmanh/websearch-crawler plus FlareSolverr) reachable at API_URL; no built-in authentication between the MCP server and the crawler. The MCP server itself takes no API key. Also installable via Smithery.
WebSearch MCP FAQ
Does this need an API key?
No. The MCP server itself takes no API key; it talks to your self-hosted crawler service at API_URL. You are responsible for running and securing that crawler backend.
What backend does it require?
A WebSearch Crawler API service, provided as the Docker image laituanmanh/websearch-crawler, alongside a FlareSolverr container. The README includes a docker-compose.yml (with an Apple Silicon workaround) to start both.
How do I install it?
Run it with `npx websearch-mcp`, install globally with `npm install -g websearch-mcp`, or install for Claude Desktop automatically via Smithery (`npx -y @smithery/cli install @mnhlt/WebSearch-MCP --client claude`).
Which clients are supported?
Any MCP client using stdio transport. The README documents configuration for Claude Desktop, Cursor IDE, and Cline, plus a Windows cmd /c workaround.
Alternatives to WebSearch MCP
Compare all alternatives →
Official MCP reference server that fetches a URL and returns its content as clean Markdown, with chunking.
Official Firecrawl MCP server — scrape, crawl, map, search, and structured extraction for any LLM client.
Exa's neural web search and crawling MCP server — runs locally via npx or as a hosted remote endpoint.
Compare WebSearch MCP with: