Skip to content

Firecrawl MCP Server

by FirecrawlConnectorResearch something

Lets your AI agent search the web and turn pages and documents into clean, readable text

View on GitHub

About Firecrawl MCP Server

Firecrawl's connector that gives your AI agent web tools: search the web, read pages as clean text, and parse PDFs and other files. With a Firecrawl account it can also crawl whole sites and watch pages for changes.

For ResearchersWritersPeople gathering web material

Made by Firecrawl

Firecrawl is a San Francisco company founded in 2022 through Y Combinator. It builds tools that turn web content into clean data for AI agents, including search, scraping and an interactive browser.

Stars
★ 7.7k
License
MIT
Cost
Free
Level
Beginner
Runs
Cloud

Overview

Firecrawl MCP Server is Firecrawl's connector that gives your AI agent web tools. Your agent can search the web, read any page as clean Markdown and parse PDFs and other files you give it; with a Firecrawl account it can also crawl a whole site, list its pages, click through pages in a browser or watch a page for changes. The hosted version needs only a web address in your agent's settings. It suits researchers, writers and anyone who wants an agent to gather material from the web.

Search, page reading and document parsing work with no account or key, and signing in later unlocks site crawls and change alerts.

Try these first

Start with the first one: it is a small task you can check easily.

  1. Read [public URL] and summarise it in five bullet points.First try · no account needed

  2. Search for recent articles on [topic] and give me three sources with their dates.

  3. Parse the PDF file [file path] and list the main figures it reports.

  4. Crawl the docs at [site], at most 20 pages, and list each page's title.needs a key, credits

Example prompts written by Agent Flows Hub.

Capabilities

From the creator's documentation.

ToolsetWhat your agent can doAccess
Web searchSearch the web, plus indexes for developer, government and science sourcesRead
Page readingTurn any web page URL into clean Markdown or structured JSONRead
Document parsingRead PDFs, Word files and other documents you give itRead
Crawl and mapCollect many pages from one site, or list its URLs; needs an accountRead
Browser actionsClick, type and fill in forms in a remote browser; needs an accountRead + write
MonitoringCheck pages on a schedule and report changes; needs an accountRead

Good to know

Give an MCP client current web material: search results, a page or document as Markdown, or many pages from one site. The agent can then summarise or compare sources it actually read.

Limits

  • Without an account the hosted server offers only search, page reading and document parsing, within daily request and credit caps per IP address. Crawls, site maps, research runs and monitors need an account or API key and use credits.
  • Retrieved pages can be out of date or wrong. Check the sources and dates before relying on a result.

Notes by (original project name: firecrawl-mcp-server), written from the creator's documentation, not from running it ourselves. Firecrawl MCP: get started · Firecrawl MCP Server: README and tool list

Stuck setting it up?

Tell us what happened. We take a look first; if it needs paid work, you get an estimate before anything starts.

Get help

Other connectors

  • Fireflies MCP Server Lets your AI assistant search your recorded meetings, summaries and action items
  • Fli Lets your AI assistant search Google Flights and find the cheapest dates to fly
  • Gamma MCP Lets your AI assistant create, edit and export Gamma presentations, docs and webpages
  • GitDiagram Turns any public GitHub project into a clickable diagram of how its parts fit together
  • GitHub MCP Server Lets your AI agent read code, manage issues and pull requests, and check builds on GitHub
  • GitMCP Gives your AI assistant the current docs and code of any GitHub project from one URL

See all 52 connectors · An MCP server is missing or disconnected

GitHub