Sponsor DartzDabz: $49/week

Home › Guides › Firecrawl MCP server

Firecrawl MCP

Search and scrape the web from your agent

Get new agent tools by email, free

Firecrawl MCP server: setup for Claude Code, Cursor and more

Updated

This page has affiliate links, each marked "Affiliate link, #ad". We may be paid if you sign up, at no extra cost to you. See our affiliate disclosure.

The Firecrawl MCP server lets an AI agent search the web, scrape pages into Markdown and read JavaScript-heavy sites. Point your MCP client (see what an MCP server is) at Firecrawl's sign-in URL, or use the API-key URL with a bearer token. Here is the setup for Claude Code, Cursor, Codex and ChatGPT, from Firecrawl's own docs.

Get new AI agent tools and setups you can copy, free.

Plus one rising trend and one agent build you can copy.

We check 2,246 agent tools daily. See a sample (count as of Oct 25, 2026)

What does the Firecrawl MCP server do?

It gives an AI agent Firecrawl's web tools inside your MCP client. Your agent can search the web, scrape a page into Markdown and read sites that need JavaScript, without you writing request code. Firecrawl runs the server for you, so you add one URL to your client and sign in. Get a Firecrawl account. (Affiliate link, #ad: we may be paid if you sign up, at no extra cost to you.)

Is the Firecrawl MCP server official?

Yes. Firecrawl documents and hosts it at mcp.firecrawl.dev, and its docs also explain how to run the open-source server locally over stdio or Streamable HTTP. Firecrawl's core is open source too. If you see other Firecrawl MCP packages, check that the URL you add is the one in Firecrawl's own docs.

Related: Firecrawl in the MCP directory · remote MCP servers trend

Sources (3) · checked Oct 25, 2026
  1. Firecrawl docs: MCP server, Get Started (OAuth server URL, API-key URL with bearer token, native ChatGPT/Codex/Claude connectors, Copilot guide, run locally)
  2. DartzDabz /partners/ Firecrawl card (searches, scrapes and interacts with pages; Markdown by default; renders JavaScript; MCP or CLI; open-source core)
  3. Firecrawl docs: MCP server, For Humans (sign-in commands for Codex, Claude Code, Cursor/JSON clients, OpenCode, Droid; revoke in MCP settings; 'not a page to open in a browser')

What is the Firecrawl MCP server URL?

There are two. For browser sign-in, use https://mcp.firecrawl.dev/v2/mcp-oauth, which needs no API key or headers. For a key in your config, use https://mcp.firecrawl.dev/v2/mcp with an Authorization bearer header. Firecrawl says both reach the same tools on your team's plan.

Sign in (OAuth), from Firecrawl's docs:

{
  "mcpServers": {
    "firecrawl": {
      "type": "http",
      "url": "https://mcp.firecrawl.dev/v2/mcp-oauth"
    }
  }
}

API key (placeholder), from Firecrawl's docs:

{
  "mcpServers": {
    "firecrawl": {
      "type": "http",
      "url": "https://mcp.firecrawl.dev/v2/mcp",
      "headers": {
        "Authorization": "Bearer <FIRECRAWL_API_KEY>"
      }
    }
  }
}

Should you sign in or use an API key?

Sign in if a person sets up the client: the client opens a browser, you approve a Firecrawl team, and no key sits in a config file. Use an API key for agents or servers that can't open a browser. Either way, you can review and revoke connections later in Firecrawl's MCP settings. Each server costs context: see the MCP context cheat sheet.

Sources (3) · checked Oct 25, 2026
  1. Firecrawl docs: MCP server, Get Started (OAuth server URL, API-key URL with bearer token, native ChatGPT/Codex/Claude connectors, Copilot guide, run locally)
  2. Firecrawl docs: MCP server, For Agents / Add an API key (bearer token JSON config on https://mcp.firecrawl.dev/v2/mcp)
  3. Firecrawl docs: MCP server, For Humans (sign-in commands for Codex, Claude Code, Cursor/JSON clients, OpenCode, Droid; revoke in MCP settings; 'not a page to open in a browser')

How do you set up the Firecrawl MCP server in Claude Code?

Add Firecrawl as an HTTP server with one command, then sign in. Firecrawl's docs give the command below. After you run it, open Claude Code, type /mcp and finish the browser sign-in. If no Firecrawl tools appear, restart Claude Code after saving the configuration.

  1. Add the server. Run: claude mcp add --transport http firecrawl https://mcp.firecrawl.dev/v2/mcp-oauth
  2. Sign in. In Claude Code, run /mcp, pick firecrawl and complete the browser sign-in. Approve the Firecrawl team you want to use.
  3. Check the tools. If no Firecrawl tools appear, restart or reload Claude Code so it reads the new configuration.
  4. Try a prompt. Ask: 'Search the web for the latest Firecrawl release notes and summarize the sources.' That is the test prompt in Firecrawl's docs.

Claude Code, from Firecrawl's docs:

claude mcp add --transport http firecrawl https://mcp.firecrawl.dev/v2/mcp-oauth

Try Firecrawl Affiliate link, #ad: we may be paid if you sign up, at no extra cost to you.

What if you use Claude Desktop instead?

Firecrawl has a separate Claude Desktop guide that connects once under the server name firecrawl, and that setup also covers Cowork and the Code tab. On Claude.ai you can skip config files: install the native Firecrawl connector from Claude's connector directory and approve access when it asks.

Related: best MCP servers for Claude Code

Sources (2) · checked Oct 25, 2026
  1. Firecrawl docs: MCP server, For Humans (sign-in commands for Codex, Claude Code, Cursor/JSON clients, OpenCode, Droid; revoke in MCP settings; 'not a page to open in a browser')
  2. Firecrawl docs: MCP server, Get Started (OAuth server URL, API-key URL with bearer token, native ChatGPT/Codex/Claude connectors, Copilot guide, run locally)

How do you add Firecrawl MCP to Cursor, Codex and ChatGPT?

Cursor takes the OAuth URL as a url entry under mcpServers in its JSON config. Codex adds it with codex mcp add, then codex mcp login. ChatGPT and Claude.ai use Firecrawl's native connector instead of a URL. Firecrawl also has a guide for GitHub Copilot CLI and VS Code.

Cursor or any JSON-config client, from Firecrawl's docs:

{"mcpServers": {"firecrawl": {"url": "https://mcp.firecrawl.dev/v2/mcp-oauth"}}}

Codex, from Firecrawl's docs:

codex mcp add firecrawl --url https://mcp.firecrawl.dev/v2/mcp-oauth
codex mcp login firecrawl

Why does the server URL show an error in my browser?

Because it is not a web page. Firecrawl's docs say the URL is for your MCP client, which starts the browser sign-in for you. Paste it into your client's config or command, not the address bar. A person still has to finish the sign-in and approve a team the first time.

Related: best MCP servers for Cursor

Sources (2) · checked Oct 25, 2026
  1. Firecrawl docs: MCP server, For Humans (sign-in commands for Codex, Claude Code, Cursor/JSON clients, OpenCode, Droid; revoke in MCP settings; 'not a page to open in a browser')
  2. Firecrawl docs: MCP server, Get Started (OAuth server URL, API-key URL with bearer token, native ChatGPT/Codex/Claude connectors, Copilot guide, run locally)

What can your agent do once Firecrawl is connected?

Ask in plain English. Your agent can search the web and summarize sources, turn a page into clean Markdown, read pages that only render with JavaScript and interact with pages. Firecrawl's docs list which tools each connection type gets, so check that page if a tool you expect is missing.

Is Firecrawl the right scraper for every job?

Not always. Our comparison guide covers when ScraperAPI is the better fit for pages that block plain requests, and when Scrapingdog's Google Search and e-commerce endpoints save work. (Affiliate link, #ad: we may be paid if you sign up, at no extra cost to you.) Firecrawl suits search plus Markdown for an LLM. Pick by the job your agent does most, not by the tool list.

Related: ScraperAPI vs Scrapingdog vs Firecrawl · ScraperAPI MCP server setup

Related: best web scraping API for AI agents · Apify MCP server

Sources (3) · checked Oct 25, 2026
  1. DartzDabz /partners/ Firecrawl card (searches, scrapes and interacts with pages; Markdown by default; renders JavaScript; MCP or CLI; open-source core)
  2. Firecrawl docs: MCP server, For Humans (sign-in commands for Codex, Claude Code, Cursor/JSON clients, OpenCode, Droid; revoke in MCP settings; 'not a page to open in a browser')
  3. DartzDabz guide: ScraperAPI vs Scrapingdog vs Firecrawl (MCP and SDK section)

See how we run our own agent crew in the playbook.

Building on Grok Bot? Start a crew from the DartzDabz Crew Builder template.

We may be paid by X based on how many people use our Crew Builder template.

What else do people ask?

What is the Firecrawl MCP server URL?

https://mcp.firecrawl.dev/v2/mcp-oauth for browser sign-in, or https://mcp.firecrawl.dev/v2/mcp with an Authorization bearer header if you use an API key.

Does the Firecrawl MCP server need an API key?

Not if you sign in. The OAuth URL opens a browser sign-in and needs no key or headers. Use a key only for clients that can't open a browser.

How do I add Firecrawl MCP to Claude Code?

Run claude mcp add --transport http firecrawl https://mcp.firecrawl.dev/v2/mcp-oauth, then run /mcp in Claude Code and finish the sign-in.

Does Firecrawl MCP work in Cursor?

Yes. Add the OAuth URL as a url entry under mcpServers in Cursor's MCP JSON config, then sign in when Cursor asks.

Is the Firecrawl MCP server open source?

Firecrawl's docs explain how to run the open-source server locally over stdio or Streamable HTTP, and Firecrawl's core is open source.

Can I use Firecrawl in ChatGPT or Claude.ai without config files?

Yes. Install Firecrawl's native connector in ChatGPT or Claude.ai and approve access.

How much does Firecrawl cost?

Plans and limits: see vendor site. We don't list affiliate vendors' prices.

See real agent setups you can copy, browse the agent tool directory and MCP servers by trust score, or submit a tool; listings are free.

Get the free DartzDabz newsletter

Free email with new and notable agent tools, the biggest trust-score changes, a rising trend, and one setup you can copy. No spam; unsubscribe anytime.