Matrix / Scraping & Extraction / Firecrawl

Firecrawl MCP server

Scraping & Extraction · remote · Up · auth · Docs + endpoint checked 2026-10-16

Firecrawl has an official hosted MCP server at https://mcp.firecrawl.dev/v2/mcp-oauth. It uses OAuth (browser sign-in) at /v2/mcp-oauth; API key (Bearer) or keyless (rate-limited, Search/Scrape/Parse only) at https://mcp.firecrawl.dev/v2/mcp over Streamable HTTP (also local stdio via npx firecrawl-mcp). Agents can use: scrape (incl. JSON structured extraction), search, map. Last checked 2026-10-16 10:06 PT: Up · auth.

Works in: ClaudeChatGPTClaude CodeCursor
Sources: Claude · ChatGPT · Claude Code · Cursor, checked 2026-10-16
Access: Read + write
Credits: Every call uses credits
Plan: Free
Setup: Self-serve
No admin step in the vendor's setup notes. Source: docs.firecrawl.dev, checked 2026-10-16
Useful for: SDRs, Marketing, RevOps

Firecrawl MCP endpoint and transport

https://mcp.firecrawl.dev/v2/mcp-oauth

Transport: Streamable HTTP (also local stdio via npx firecrawl-mcp)
Docs: Official Firecrawl MCP docs

Connect Firecrawl without code

Firecrawl's docs confirm it works in Claude and ChatGPT. You don't need a terminal or a config file. Check Firecrawl's docs first: some vendors give you a personal URL or an API key to use in place of the plain endpoint.

Claude (web and desktop app)

If Firecrawl shows up when you browse connectors in Claude, connect it from there instead. That skips steps 2 to 4.

In Claude, open Customize → Connectors.
Click + Add, then Add custom connector.
Name it “Firecrawl”, paste the server URL https://mcp.firecrawl.dev/v2/mcp-oauth, and click Continue.
Review the sign-in settings Claude detected and click Add.
Click Connect and sign in to Firecrawl. Then ask Claude something like “What can the Firecrawl connector do?”

On Team and Enterprise plans, an Owner adds the connector first under Organization settings → Connectors → Add → Custom → Web. Each person then connects it from Customize → Connectors. Free plans can add one custom connector.

Steps from the Claude Help Center: support.claude.com, checked 2026-10-16

ChatGPT

If Firecrawl has its own plugin in ChatGPT's directory, install that instead.

Go to chatgpt.com/plugins.
Select +, then Add custom MCP server. If you don't see it, your plan or workspace settings may not allow custom servers, so ask your workspace admin.
Enter a name and description. Under Connection, paste the server URL https://mcp.firecrawl.dev/v2/mcp-oauth.
Set up sign-in, read the warning, and select I understand and want to continue.
Select Create as a plugin. In a new chat, type @ and pick Firecrawl.

Steps from the OpenAI developer docs: developers.openai.com, checked 2026-10-16

From Firecrawl's docs: ChatGPT and Claude.ai use the native Firecrawl connector; Claude Code, Cursor and Codex add the server URL (OAuth sign-in at /v2/mcp-oauth or API key Bearer at /v2/mcp). Keyless connections expose only Search, Scrape and Parse.Source: docs.firecrawl.dev, checked 2026-10-16

Developer setup (Claude Code, Cursor)

Add this to .mcp.json (Claude Code, project root) or ~/.cursor/mcp.json (Cursor). Remove the // comment line if your client requires strict JSON.

{
  "mcpServers": {
    "firecrawl": {
      "url": "https://mcp.firecrawl.dev/v2/mcp-oauth"
      // headless alternative: "headers": { "Authorization": "Bearer YOUR_FIRECRAWL_API_KEY" }
    }
  }
}

Add the endpoint to your client config (above), or as a custom connector in Claude.
Complete the OAuth sign-in. In Claude Code, run /mcp and pick the server.
Test it with a prompt like: “List the tools the Firecrawl MCP server exposes.”

Firecrawl MCP authentication

OAuth (browser sign-in) at /v2/mcp-oauth; API key (Bearer) or keyless (rate-limited, Search/Scrape/Parse only) at https://mcp.firecrawl.dev/v2/mcp

What it can change

Read + write Some Firecrawl MCP tools change data, not just read it.

Things it can do in your Firecrawl account:

creates, updates and deletes page-change monitors
operates live web pages (clicks, form fills, navigation)
submits product feedback to Firecrawl

Source: docs.firecrawl.dev, checked 2026-10-16

Review each change before you approve it, especially anything that sends messages or edits records other people rely on.

Cost and credits

Every call uses credits

Firecrawl is credit-based: Scrape 1 credit/page, Crawl 1 credit/page, Map 1 credit/call, Search 2 credits/10 results, Interact 2-7 credits/browser minute, plus option add-ons (e.g. JSON extraction +4/page). Alexandria runs cost the price listed on each tool; discovery is free. Keyless MCP is free but capped per IP per day by requests and credits.

Source: docs.firecrawl.dev, checked 2026-10-16

Plan required

Free

Source: docs.firecrawl.dev, checked 2026-10-16

Admin and governance

Controls Firecrawl itself provides. Settings inside Claude or ChatGPT aren't counted here.

An admin must approve or enable the connectionNot confirmed
Scopes or permissions can limit what it reachesNot confirmed
Individual tools can be turned offNot confirmed
MCP activity shows in an audit logNot confirmed

What the Firecrawl MCP server can do

scrape (incl. JSON structured extraction)
search
map
crawl
parse files
agent (autonomous research)
interact (live browser: clicks, form fills)
monitors (page-change tracking)
Alexandria third-party data tools
research/developer/government search indexes

Firecrawl MCP status

Up · auth HTTP 401 · 238 ms · checked 2026-10-16 10:06 PT. Full status and history.

Recipes that use Firecrawl

Competitor pricing-page watch (Firecrawl + Notion)

Firecrawl MCP FAQ

What is the Firecrawl MCP server URL?

https://mcp.firecrawl.dev/v2/mcp-oauth

How does Firecrawl MCP authenticate?

OAuth (browser sign-in) at /v2/mcp-oauth; API key (Bearer) or keyless (rate-limited, Search/Scrape/Parse only) at https://mcp.firecrawl.dev/v2/mcp.

Which apps does Firecrawl MCP work with?

Firecrawl's docs confirm it works in Claude, ChatGPT, Claude Code and Cursor. Other apps aren't confirmed in the docs yet.

Can I use Firecrawl in Claude or ChatGPT without code?

Yes. Firecrawl's docs confirm it works in Claude and ChatGPT. Add it as a connector in the app settings and sign in. No terminal or config file needed.

Can the Firecrawl MCP server change my data?

Yes. Some tools change data, for example: creates, updates and deletes page-change monitors, operates live web pages (clicks, form fills, navigation), submits product feedback to Firecrawl. Review each change before you approve it.

Does Firecrawl MCP use credits?

Every call uses credits. Firecrawl is credit-based: Scrape 1 credit/page, Crawl 1 credit/page, Map 1 credit/call, Search 2 credits/10 results, Interact 2-7 credits/browser minute, plus option add-ons (e.g. JSON extraction +4/page). Alexandria runs cost the price listed on each tool; discovery is free. Keyless MCP is free but capped per IP per day by requests and credits.

What Firecrawl plan do I need for MCP?

Free.

Is the Firecrawl MCP server official?

Yes. It's documented on docs.firecrawl.dev, last checked 2026-10-16.

Related MCP servers

Apify (Scraping & Extraction)
Exa (AI Research)
Perplexity (AI Research)
Tavily (AI Research)
Wynter (Market Research)

Something out of date on this page? Report a correction, with a link to the vendor page that shows the right information.

Firecrawl MCP server

Firecrawl MCP endpoint and transport

Connect Firecrawl without code

Claude (web and desktop app)

ChatGPT

Developer setup (Claude Code, Cursor)

Firecrawl MCP authentication

Get told when Firecrawl MCP changes

What it can change

Cost and credits

Plan required

Admin and governance

What the Firecrawl MCP server can do

Firecrawl MCP status

Recipes that use Firecrawl

Firecrawl MCP FAQ

What is the Firecrawl MCP server URL?

How does Firecrawl MCP authenticate?

Which apps does Firecrawl MCP work with?

Can I use Firecrawl in Claude or ChatGPT without code?

Can the Firecrawl MCP server change my data?

Does Firecrawl MCP use credits?

What Firecrawl plan do I need for MCP?

Is the Firecrawl MCP server official?

Related MCP servers