A Model Context Protocol (MCP) server implementation that integrates with Firecrawl to provide comprehensive web scraping capabilities. This MCP server connects Firecrawl’s scraping, crawling, discovery, search, and data extraction features to MCP clients, enabling automated, scriptable web data workflows. Supported tools include scrape, batch_scrape, map, search, crawl, and extract, covering single-page extraction, multi-page batch processing, URL discovery, and structured data extraction. It offers automatic retries with exponential backoff and built-in rate limiting to handle transient errors and API constraints. It supports both cloud API usage (requiring a FIRECRAWL_API_KEY) and self-hosted deployments via FIRECRAWL_API_URL, and can run in various environments such as Cursor, Windsurf, or Streamable HTTP Local Mode. Configuration options cover retry policies and credit usage monitoring, along with rate limiting controls and SSE streaming support for real-time updates. The server is designed for cloud and self-hosted setups, enabling deep scraping workflows, batch research, and scalable processing. Example configurations and integration notes are provided to assist Composer Agents and MCP clients in building advanced web-scraping workflows.