This document describes a Model Context Protocol (MCP) server implementation using Firecrawl for web scraping. The server offers features like web scraping with JS rendering, URL crawling, content extraction, and batch scraping. It supports both cloud and self-hosted Firecrawl instances and integrates with Cursor, Windsurf, and Claude Desktop. Installation instructions are provided for each platform, including configuration details for API keys and optional parameters. Several tools are available, including scrape, batch scrape, search, crawl, extract, deep research, and llms.txt generation tools. Each tool's usage and parameters are explained with examples. The server incorporates retry logic, credit usage monitoring, and rate limiting for robust operation. Comprehensive logging and error handling mechanisms are also included. Configuration examples demonstrate how to customize retry behavior, credit thresholds, and API endpoints. The server aims to provide a reliable and efficient solution for diverse web scraping tasks.