Firecrawl Mcp Server

What is Firecrawl MCP Server?

Firecrawl MCP Server is a Model Context Protocol server that brings Firecrawl’s web scraping, searching, and interactive browsing capabilities to MCP-compatible AI agents. It is designed for developers and AI agent builders who need clean, structured live web data for their applications.

How to use Firecrawl MCP Server?

The simplest way is to connect to the hosted server at https://mcp.firecrawl.dev/v2/mcp (keyless free tier available). Alternatively, run env FIRECRAWL_API_KEY=fc-YOUR_API_KEY npx -y firecrawl-mcp from the command line. Configure the server by setting the FIRECRAWL_API_KEY environment variable, and optionally FIRECRAWL_API_URL for self-hosted instances or FIRECRAWL_OAUTH_TOKEN for OAuth bearer tokens.

Key features of Firecrawl MCP Server

• Search the web and get full page content
• Scrape any URL into clean, structured data
• Interact with pages via click, navigate, and operate
• Deep research with an autonomous agent
• Automatic retries and rate limiting
• Cloud and self-hosted deployment support
• SSE transport support

Use cases of Firecrawl MCP Server

• Extracting single-page content from a known URL using structured JSON
• Discovering all URLs on a website for site mapping
• Complex multi-source research across unknown sources via the agent tool
• Interactive browser automation such as clicking, typing, and navigating
• Structured extraction from URLs using the extract tool

FAQ from Firecrawl MCP Server

What tools are available in Firecrawl MCP Server?

The server provides firecrawl_scrape for single-page extraction, firecrawl_search for web search, firecrawl_interact for browser automation, firecrawl_map for URL discovery, firecrawl_crawl for multi-page extraction, firecrawl_extract for structured data, firecrawl_agent for complex research, and firecrawl_monitor for recurring page checks.

When should I use JSON format versus Markdown for scraping?

Use JSON format with a schema to extract only the specific data you need—this keeps responses small and avoids context window overflow. Use Markdown format only when you genuinely need the full page content, such as reading an entire article for summarization.

Do I need an API key to use Firecrawl MCP Server?

On the keyless free tier, the scrape, search, and interact tools work without an API key (though rate-limited). Other tools such as crawl, map, agent, and extract require a Firecrawl API key. You can also use OAuth bearer tokens (starting with fco_).

Can I run Firecrawl MCP Server with a self-hosted Firecrawl instance?

Yes. Set the FIRECRAWL_API_URL environment variable to your custom API endpoint (e.g., https://firecrawl.your-domain.com). Authentication via FIRECRAWL_API_KEY is optional for self-hosted instances.

How do I connect Firecrawl MCP Server to different AI clients?

You can connect to Claude Desktop, Cursor, Windsurf, VS Code, or any MCP-compatible client. The README provides specific configuration JSON blocks for each client, pointing to either the hosted endpoint or a local npx command.