#7MCP server

Firecrawl MCP

Turns web pages and whole sites into markdown an agent can read.

Save Visit website ↗

Rank

of 445 · by weekly downloads

Weekly downloads

191K

npm + PyPI, last 7 days

Add it to your client

Official

Install

npx -y firecrawl-mcp

mcpServers

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

Transport: both
Runtime: Node
Auth: Required
License: MIT

Tools

firecrawl_scrapefirecrawl_searchfirecrawl_mapfirecrawl_crawlfirecrawl_parsefirecrawl_agentfirecrawl_interact

Repositoryhttps://github.com/firecrawl/firecrawl-mcp-server

Firecrawl hands web pages to an agent as clean markdown, or as JSON that matches a schema you pass, instead of raw HTML. The agent can scrape a known URL, list a site's URLs, crawl a section of a site, or run a web search and fetch the result pages in the same call.

Setup

Hosted: https://mcp.firecrawl.dev/v2/mcp. Without a key, only firecrawl_scrape, firecrawl_search and firecrawl_parse work, with rate limits. For the full tool set, send Authorization: Bearer <FIRECRAWL_API_KEY> from the client's header settings or sign in through https://mcp.firecrawl.dev/v2/mcp-oauth.
Local: env FIRECRAWL_API_KEY=fc-YOUR_API_KEY npx -y firecrawl-mcp. FIRECRAWL_API_URL points it at a self-hosted Firecrawl, where the key is optional.

What it exposes

firecrawl_scrape, firecrawl_map, and firecrawl_crawl, which starts a job and polls until it finishes
firecrawl_search; add scrapeOptions to get page content with the results
firecrawl_interact for pages that need a click or typing before they can be read
firecrawl_agent for multi-site research, checked with firecrawl_agent_status
firecrawl_parse for PDFs, Word files and spreadsheets

Limits

Crawl results can overflow the model's context; bound them with limit and maxDiscoveryDepth.
Calls use Firecrawl credits; firecrawl_credit_usage shows what is left.

Suited to research and data-collection tasks that span many pages or several sites.

Comments(0)

No comments yet — be the first.

Similar MCP servers

AgentQL MCP Server

MCP server

Extracts structured data from any page by describing the shape you want, not the selectors.

Browsers & AutomationSearch Engines & Web

Screaming Frog MCP (headless)

MCP server

Runs Screaming Frog crawls with the app closed, through nine read-and-export tools and nothing else.

Data & AnalyticsSearch Engines & Web

Exa MCP

MCP server

Search for agents, where a result carries page content rather than a link.

Search Engines & Web