Firecrawl MCP Server

Content

Firecrawl MCP exposes Firecrawl web data tools through Model Context Protocol. The official documentation covers a hosted endpoint and local npx setup for Claude Desktop, Claude Code, Cursor, VS Code, Windsurf, Google Antigravity, and n8n. Typical tools include scraping, crawling, mapping, searching, and structured extraction.

Installation

For Claude Code local stdio setup:

claude mcp add firecrawl -e FIRECRAWL_API_KEY=your-api-key -- npx -y firecrawl-mcp

For hosted HTTP setup, use Firecrawl's documented endpoint pattern and pass the API key through the client-supported auth mechanism. For local JSON config:

{
  "mcpServers": {
    "firecrawl": {
      "command": "npx",
      "args": ["-y", "firecrawl-mcp"],
      "env": {
        "FIRECRAWL_API_KEY": "YOUR_API_KEY"
      }
    }
  }
}

Use Cases

• Scrape a public documentation page and summarize key implementation details.
• Crawl a small website section before drafting migration notes.
• Map available URLs before deciding what to fetch.
• Extract structured fields from pages where a normal browser search is too shallow.
• Give an agent web research tools while keeping API credentials in local config.

Safety and Privacy

Firecrawl is powerful enough to collect more data than intended. Use it with explicit domains, bounded page counts, and prompts that forbid authenticated or private pages unless that access is intentional. Keep API keys in environment variables or client secret storage.

Duplicate Check

content/tools/firecrawl.mdx lists Firecrawl as a tool. This MCP entry is a separate installable integration for MCP clients and belongs in content/mcp.

References

• Firecrawl MCP docs - https://github.com/firecrawl/firecrawl-docs/blob/main/mcp-server.mdx
• Firecrawl API keys - https://firecrawl.dev/app/api-keys
• Firecrawl hosted MCP endpoint - https://mcp.firecrawl.dev/v2/mcp