# Firecrawl MCP Server

The official Firecrawl MCP server, giving MCP-compatible AI agents web search, scraping, page interaction and deep research over the live web as clean, agent-ready context.

Firecrawl MCP Server is a Model Context Protocol server that brings Firecrawl's web scraping and search to MCP-compatible AI agents such as Cursor and Claude. It can search the web and return full page content, search a code-oriented index covering GitHub issues, merged pull requests, READMEs and documentation, scrape a URL into clean Markdown or JSON matching a supplied schema, discover URLs on a site without fetching their content, and crawl content from many pages under a site with limits and path filters.

A dedicated interact tool clicks, navigates and operates on a page, running one prompt or code turn to completion per call, with a session that can persist across calls and be explicitly stopped. Monitor tools check a recurring page on a schedule and report diffs, and a credit-usage tool reports current and historical Firecrawl credit consumption, optionally broken down by API key.

It supports both cloud and self-hosted deployment, includes automatic retries and rate limiting, and lists 26 tools in its default configuration, with feedback tools and search options that can be disabled through environment variables.

Website: https://github.com/firecrawl/firecrawl-mcp-server