Hosting

Remote hosted or local; cloud API or self-hosted Firecrawl API

Transport

Streamable HTTP (hosted/local); stdio (local); SSE support documented in repository

Endpoint

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

Authentication

Hosted OAuth via browser, API key using https://mcp.firecrawl.dev/v2/mcp with Authorization: Bearer <FIRECRAWL_API_KEY>, or limited keyless access. Tool availability and limits depend on plan/team policy.

Requirements

• Compatible MCP client
• Firecrawl account/API key for full authenticated access; keyless hosted use is limited to Search, Scrape, and Parse within daily limits
• Node.js/npm for local npx installation, or Docker for container deployment

Setup steps

1. For interactive hosted OAuth, configure https://mcp.firecrawl.dev/v2/mcp-oauth in the client and complete sign-in/team consent in the browser.
2. For keyless use, connect to https://mcp.firecrawl.dev/v2/mcp; only firecrawl_search, firecrawl_scrape, and firecrawl_parse are available with daily limits.
3. For API-key access, use https://mcp.firecrawl.dev/v2/mcp and store FIRECRAWL_API_KEY in the client secret/header settings, never in chat or a URL.
4. For local stdio, install/run npx -y firecrawl-mcp and set FIRECRAWL_API_KEY; local mode can optionally point FIRECRAWL_API_URL at a self-hosted API.
5. Reload/reconnect the client and inspect its tools list; available tools vary by connection mode, plan, configuration, and team policy.

Removed the stale fixed count 27; retained 12 documented rows as a partial snapshot because exact available tools vary by connection and deployment.

• firecrawl_scrape

  Extract content or structured fields from a single known URL.

  01
• firecrawl_map

  Discover URLs indexed on a website.

  02
• firecrawl_search

  Search the web and optionally fetch content from results.

  03
• firecrawl_parse

  Parse files into agent-ready content.

  04
• firecrawl_crawl

  Crawl a site or section with bounded paths/depth/limit.

  05
• firecrawl_check_crawl_status

  Resume/check a crawl job after a wait timeout.

  06
• firecrawl_agent

  Run asynchronous multi-source research.

  07
• firecrawl_agent_status

  Check status of an agent research job.

  08
• firecrawl_interact

  Click, navigate, or interact with a live page.

  09
• firecrawl_interact_stop

  Close an Interact session.

  10
• firecrawl_find_tools

  Browse the Alexandria third-party data tool catalog.

  11
• firecrawl_credit_usage

  Inspect current or historical Firecrawl credit usage.

  12

This server’s available tools can vary by account, permissions, or configuration. The list shown is not its full runtime catalog.

• Hosted keyless mode exposes only Search, Scrape, and Parse and has daily limits.
• The full tool set depends on plan, organization policy, environment flags, and enabled self-hosted services.
• Hosted MCP cannot directly read files on the user's machine; local parsing uses an upload handoff.
• Some monitoring tools create recurring checks; deletion is permanent.
• Tool call credits and rate limits vary; consult current plan and rate-limit documentation.

• official tool reference

  Official default profile documents 27 tools; optional feedback features, hosted keyless mode, plan/team policy, and deployment configuration alter exposure. The stored names are a subset.

• official documentation

  Official default profile documents 27 tools; optional feedback features, hosted keyless mode, plan/team policy, and deployment configuration alter exposure. The stored names are a subset.

• official vendor documentation

  Official docs say tool surface depends on hosted OAuth/API key/keyless/local mode, plan, team policy and deployment; keyless currently exposes three tools.

Documentation checked 2026-10-28T00:00:00Z. Connection has not been runtime tested.