Firecrawl MCP Server

by firecrawl

Searches, scrapes, crawls, maps, parses, and monitors web content through Firecrawl MCP tools.

Search & Data Extraction
7.7kstars
No ratings
1installs
UpdatedOct 27, 2026

InstallRepository Website

Install Firecrawl MCP Server

Pick your client, then add it in one click or paste the config.

STDIONode.js

~/Library/Application Support/Claude/claude_desktop_config.json

claude_desktop_config.json

{
  "mcpServers": {
    "firecrawl-firecrawl-mcp-server": {
      "command": "npx",
      "args": [
        "-y",
        "firecrawl-mcp"
      ],
      "env": {
        "FIRECRAWL_API_KEY": "YOUR_VALUE_HERE",
        "HTTP_STREAMABLE_SERVER": "YOUR_VALUE_HERE"
      }
    }
  }
}

Settings → Developer → Edit Config, merge this into mcpServers, then fully quit and reopen Claude Desktop.

Values are written into the snippet above so you can paste it as is. They never leave your browser.

Install not yet sandbox-tested— the server itself is confirmed live (we called its tools/list endpoint). The stdio install command is a separate, ongoing check.

Overview

firecrawl/firecrawl-mcp-server connects MCP clients to Firecrawl for web search, page extraction, crawling, document parsing, browser interaction, and structured research. It supports hosted MCP over HTTP with OAuth or API-key authentication, as well as local npm, Docker, and self-hosted deployments. Reach for it when an agent needs content from known URLs, URL discovery, multi-page collection, rendered-page interaction, or asynchronous multi-source research. Results can be returned as Markdown, HTML, links, screenshots, metadata, targeted answers, or schema-conforming JSON.

Use cases

Search the web for ranked sources and scrape selected pages
Discover a site's URLs before collecting page content
Crawl bounded sections of a documentation site
Extract structured fields from known pages
Track recurring page or search changes

Key features

Web, developer, GitHub, and research search
Markdown, HTML, links, screenshots, and JSON extraction
Site mapping and bounded multi-page crawling
PDF and document parsing
Live browser interaction
Asynchronous research and change monitoring

Tools25

What Firecrawl MCP Server lets your agent do. Expand a tool for its parameters.

Verified live~8.1k tokens of context

Security & health

What it can reach and how reliably it runs — from our own checks, not the maintainer’s claims.

ActiveLast checked Today

Recent health check succeeded.

33/33 checks healthy over the last 79d

Supply-chain signal

No high-severity advisories surfaced by our automated scan.

Critical 0High 0Medium 0Low 0

Scanned 8d ago via OSV.dev · firecrawl-mcp (npm)

Adoption & maintenance

Factual signals from GitHub, npm, and our automated checks — not a rating.

GitHub stars7.7kStargazers on the source repository.

npm downloads541kPackage downloads in the last 30 days.

Last commitTodayMost recent push to the default branch.

Tools exposed25Callable tools this server registers over MCP.

Directory activity1 installsConfig copies, upvotes, and views on AllMCPs.

How Firecrawl MCP Server works

What firecrawl/firecrawl-mcp-server does

firecrawl/firecrawl-mcp-server exposes Firecrawl capabilities through the Model Context Protocol. It gives compatible clients tools for searching the web, finding URLs within a site, extracting content from known pages, crawling multiple pages, parsing uploaded documents, and working with live browser sessions. It also supports structured extraction, asynchronous research jobs, recurring change monitors, developer-focused search, and scientific literature search.

The server is suited to agents that need current web evidence or machine-readable page data rather than only a browser-like text lookup. Search can return ranked web, news, or image results, while developer search covers indexed GitHub issues, pull requests, repository READMEs, and selected documentation. Research tools cover paper search, metadata, related papers, and question-focused passages from indexed full text.

How it works

Use firecrawl_search when the starting point is a query, firecrawl_map when URLs must be discovered under a site, and firecrawl_scrape when the target URL is already known. firecrawl_crawl collects content across multiple pages with controls for paths, depth, limits, subdomains, sitemaps, and scrape options. Crawls return a job identifier and status; firecrawl_check_crawl_status reads progress and results for an existing crawl.

For longer research tasks, firecrawl_agent starts an asynchronous job from a prompt and optional seed URLs. It returns a job ID, which firecrawl_agent_status polls until completion or failure. The synchronous search-and-scrape tools are a better fit when the task cannot wait several minutes.

firecrawl_interact opens or reuses a live browser session for navigation, clicks, form filling, or browser code. Sessions can be stopped with firecrawl_interact_stop. Because interaction operates on live pages, form submissions and similar actions may create external side effects.

Setup and configuration

The hosted MCP endpoint is https://mcp.firecrawl.dev/v2/mcp. Its keyless tier supports scrape, search, and parse with rate limits; other capabilities such as crawl, map, and agent require an API key. OAuth connections use https://mcp.firecrawl.dev/v2/mcp-oauth. API-key clients keep the main endpoint and send the key as a bearer secret rather than placing it in the URL.

For local use, run npx -y firecrawl-mcp with FIRECRAWL_API_KEY. The package can also be installed globally with npm, or run from the ghcr.io/firecrawl/firecrawl-mcp-server Docker image. Setting HTTP_STREAMABLE_SERVER=true starts local Streamable HTTP mode at http://localhost:3000/mcp instead of the default stdio transport.

Cursor and Windsurf configurations use an npm command with the API key supplied through the client's environment or secret configuration. The hosted search-only endpoint, https://mcp.firecrawl.dev/v2/mcp-search, provides a fixed search-oriented tool set.

Tools and capabilities

Search web, news, image, developer, GitHub, and research sources.
Scrape known URLs into Markdown, HTML, links, screenshots, metadata, answers, or schema-based JSON.
Map site URLs and crawl bounded sets of pages.
Parse HTML, PDF, Word, RTF, OpenDocument, and spreadsheet files.
Run asynchronous structured research jobs and inspect their results.
Interact with JavaScript-rendered pages through browser sessions.
Create, run, inspect, update, list, and delete recurring content monitors.

Limitations and notes

Search results with attached page content use scraping for each fetched result, while search-only requests return results without page bodies. Firecrawl may reuse recently indexed content; set scrape options such as maxAge: 0 when a live fetch is required. Crawl output can be large, so page and depth limits are important for broad sites.

The firecrawl_extract entry point is deprecated; use firecrawl_scrape with JSON output and a supplied schema instead. Hosted document parsing uses an upload step before the final parse call, while local parsing reads from the server filesystem. Browser interaction runs one prompt or code turn per call, although a session can persist through its scrape ID.

Read the full README View source on GitHub

Reviews

No reviews yet — be the first to share how it worked for you.

Frequently Asked Questions about Firecrawl MCP Server

firecrawl/firecrawl-mcp-server is an MCP server that connects compatible AI clients to Firecrawl for web search, URL scraping, site mapping, crawling, document parsing, browser interaction, structured extraction, asynchronous research, and content monitoring. It can return Markdown, HTML, links, screenshots, metadata, targeted answers, or schema-based JSON.

Install it with `npx -y firecrawl-mcp` and provide `FIRECRAWL_API_KEY` in the client environment. You can also install the package globally with npm, run the published Docker image, or connect to the hosted MCP endpoint.

Yes, it works with Cursor version 0.45.6 or newer. Add an MCP server using the `npx -y firecrawl-mcp` command and set `FIRECRAWL_API_KEY` in the configuration.

Yes, `firecrawl_crawl` collects content from multiple pages and supports limits for paths, depth, page count, subdomains, external links, sitemaps, delays, and scrape options. Use `firecrawl_map` first when you only need the site's URL inventory.

Use `firecrawl_scrape` on a known URL with JSON output and provide a prompt and schema describing the desired fields. The deprecated `firecrawl_extract` tool should be replaced by this scrape-based workflow.

Related MCP servers

All in Search & Data Extraction All alternatives

Add a badge to your READMEShow it’s listed — live SVG, Markdown or HTML.

AllMCPs Directory Badge

Full Badge Customizer

Showcase your server listing on GitHub or your project documentation. Embed this dynamic SVG badge to highlight official listing status and live engagement.

Badge Style:

Live Dynamic SVG Preview

Firecrawl MCP Server AllMCPs Directory Badge

Markdown (GitHub README)

[![AllMCPs](https://allmcps.com/api/badge/firecrawl-firecrawl-mcp-server?style=directory)](https://allmcps.com/mcp/firecrawl-firecrawl-mcp-server)

HTML Embed

<a href="https://allmcps.com/mcp/firecrawl-firecrawl-mcp-server"><img src="https://allmcps.com/api/badge/firecrawl-firecrawl-mcp-server?style=directory" alt="Firecrawl MCP Server on AllMCPs" /></a>