Firecrawl MCP
Connect Firecrawl to MCP clients, agents, and unattended workloads.
Firecrawl MCP gives AI clients access to live web data through the Model Context Protocol. It is available as a hosted service or as an open-source local server.
Start here
Choose the connection mode that matches where Firecrawl will run. The hosted routes are credential-free URLs: credentials belong in the OAuth flow or an Authorization header, never in the URL itself.
Connect an account
Use browser-based OAuth for an interactive MCP client and the full Firecrawl tool surface.
Run unattended
Use /v2/mcp with a Firecrawl API key in an Authorization header for CI, servers, and scripts.
Try keyless MCP
Start without an API key on the rate-limited Search, Scrape, and Parse surface.
Learn more
Client setup
Configure Claude, Cursor, VS Code, Windsurf, n8n, and other clients.
Run locally
Install the open-source MCP server and configure a local or self-hosted deployment.
Tools and operations
Review Firecrawl MCP tools, rate limits, logging, and error handling.
Develop Firecrawl MCP
Build, test, and contribute to the open-source server.
For the current keyless allowance and rate limits, see Rate limits. For broader agent onboarding, see Build with AI.