Firecrawl MCP Server Guide

Name: Firecrawl MCP Server
Author: Firecrawl

Overview

What Firecrawl MCP Server does

Firecrawl MCP Server brings Firecrawl's web data APIs into MCP clients. It can turn a page into clean content, map a site's URLs, search the web, crawl a domain, extract structured fields, parse documents, drive page interactions, and run longer research or monitoring jobs. Users can connect to Firecrawl's hosted MCP endpoint or run the npm package locally against the cloud API or a compatible self-hosted Firecrawl deployment. Several jobs are asynchronous, so agents receive an identifier and check status in later calls.

Best for

Research and documentation workflows that need current web content
Collecting structured information from pages that need rendering or extraction
Agents that need a consistent API for search, scrape, crawl, and browser interaction

Not ideal for

Offline environments with no reachable web or Firecrawl service
Unlimited high-volume crawling without a credit and rate-limit budget

Capabilities

What an agent can do

01
Scrape a URL into Markdown, HTML, links, screenshots, or structured data
02
Search the web and optionally attach extracted page content
03
Map and crawl sites with asynchronous status tracking
04
Parse documents and extract schema-shaped results
05
Interact with rendered pages and run agentic research or monitoring jobs

Representative tools and operations

firecrawl_scrapefirecrawl_mapfirecrawl_searchfirecrawl_crawlfirecrawl_check_crawl_statusfirecrawl_parsefirecrawl_extractfirecrawl_agentfirecrawl_agent_statusfirecrawl_interactfirecrawl_interact_stop

Installation

Connect Firecrawl MCP Server

Use the publisher’s current instructions as the source of truth. The examples below were checked on Jul 11, 2026.

Any stdio MCP clientRun locally with the Firecrawl npm package

Command

env FIRECRAWL_API_KEY=fc-YOUR_API_KEY npx -y firecrawl-mcp

Configuration

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

Remote MCP clientsUse Firecrawl's hosted endpoint

Command

https://mcp.firecrawl.dev/{FIRECRAWL_API_KEY}/v2/mcp

Configuration

Prefer an OAuth Bearer token or secret-managed API key; do not commit a key inside a client configuration file.

Trust and access

Authentication and security notes

Authentication: Use FIRECRAWL_API_KEY for the cloud API or an authenticated self-hosted endpoint. HTTP transports can accept a Firecrawl OAuth Bearer token.

Review before connecting

Treat scraped pages as untrusted input and prevent page text from changing agent policy or approving tool calls.
Store API keys outside source control and use separate keys or endpoints for production and experimentation.
Set crawl limits and review interaction or extraction targets to avoid collecting prohibited or sensitive material.

Known limitations

Cloud usage consumes Firecrawl credits and is subject to account limits.
Some document and file flows differ between hosted MCP and local self-hosted modes.
Dynamic sites, access controls, robots policies, and anti-bot measures can affect results.

Evidence

Sources used for this guide

Facts were checked against primary publisher material. Descriptions and guidance are original Graphify summaries.

FAQ

Questions about Firecrawl MCP Server

Can Firecrawl MCP work with a self-hosted Firecrawl API?

Yes. Set FIRECRAWL_API_URL to the compatible self-hosted endpoint and provide a key only if that deployment requires one.

Why do some Firecrawl tools return a job ID?

Crawls, agent research, and similar operations may take longer than one request, so the server separates job creation from status retrieval.