MCP server · Docs and web search
Firecrawl MCP server
Lets an assistant scrape, search, map and crawl websites through Firecrawl and get clean Markdown or structured JSON back.
- Maintainer
- Firecrawl (official)
- Licence
- MIT
- Runs
- Remote (URL)
- Sign-in
- Optional token
- Registry version
- 3.30.1 (updated 2026-10-15)
- Last activity
- 2026-10-13
Overview
What it does
Firecrawl’s MCP server brings its web data service to assistants. The hosted endpoint is the documented starting point: without a key it offers scrape, search and parse at rate-limited levels, and an API key in an Authorization header or an OAuth sign-in at a separate URL opens up the full set of tools. The same server runs locally through npx, and it can point at a self-hosted Firecrawl instance.
Use scrape for a known URL, as Markdown or as JSON that matches a schema you supply. Map discovers a site’s URLs, crawl collects many pages with depth and path limits, and search starts from a query. Interact clicks and types on a page before reading it, an agent tool handles deeper research, monitor tools check pages on a schedule and report changes, and a credit tool shows usage. A search-only endpoint offers a fixed set of nine search tools.
Tools
What the assistant can do
- Scrape a URL to Markdown or schema-matched JSON
- Map a site’s URLs and crawl many pages with limits
- Web search, with optional page scraping in the same call
- Interact with pages by clicking, typing and navigating
- Research agent runs and scheduled page monitoring with change alerts
- Check credit usage
Install
Install Firecrawl MCP server in your app
You need:
- An MCP client that supports remote HTTP servers
- A Firecrawl API key for the full tool set (keyless use covers scrape, search and parse)
- Node.js 22 or newer if you run the npm package locally
Run in your project. --scope local keeps the servers private to you in this project; use user for every project.
claude mcp add --scope local --transport http firecrawl https://mcp.firecrawl.dev/v2/mcp
- firecrawl: if the server needs a sign-in, the app will ask (run /mcp in Claude Code).
Project root. Commit it to share the servers with your team; each person sets the variables in their own shell.
{
"mcpServers": {
"firecrawl": {
"type": "http",
"url": "https://mcp.firecrawl.dev/v2/mcp"
}
}
}
- firecrawl: if the server needs a sign-in, the app will ask (run /mcp in Claude Code).
macOS: ~/Library/Application Support/Claude/claude_desktop_config.json · Windows: %APPDATA%\Claude\claude_desktop_config.json (Settings → Developer → Edit Config). Restart Claude after saving.
{
"mcpServers": {}
}
- firecrawl is a remote server: add it as a custom connector (Customize → Connectors → + Add → Add custom connector; on Team and Enterprise plans an owner adds it under Organization settings → Connectors), with the URL https://mcp.firecrawl.dev/v2/mcp.
In your project, or ~/.cursor/mcp.json for every project.
{
"mcpServers": {
"firecrawl": {
"url": "https://mcp.firecrawl.dev/v2/mcp"
}
}
}
- firecrawl: if the server needs a sign-in, the app will ask.
In your workspace. VS Code asks for each secret the first time a server starts and stores it securely.
{
"servers": {
"firecrawl": {
"type": "http",
"url": "https://mcp.firecrawl.dev/v2/mcp"
}
}
}
- firecrawl: if the server needs a sign-in, the app will ask.
macOS/Linux: ~/.config/devin/mcp_config.json · Windows: %APPDATA%\devin\mcp_config.json. Read by the default Devin Local agent and by the legacy Cascade agent.
{
"mcpServers": {
"firecrawl": {
"url": "https://mcp.firecrawl.dev/v2/mcp"
}
}
}
- firecrawl: if the server needs a sign-in, the app will ask.
Want it alongside other servers? Open Firecrawl MCP Server in the MCP config generator and add the rest; it writes one file per app. The snippets follow the setup in the server’s README, checked on 2026-10-14.
Before you install
- Scrape, crawl and agent calls use your Firecrawl credits; set crawl limits and path filters to keep jobs bounded.
- The README says never to put the API key in the server URL or in chat; set it as a header or client secret.
- Scraped content can contain prompt-injection text, and interact can submit forms on live sites.
- A remote server sees whatever the assistant sends it. Sign in with the narrowest access the service offers. See how to choose a safe server.
Why it’s listed
Why we list it
A widely used scraping and crawling server with a key-less hosted tier, maintained by Firecrawl with frequent releases.
Related
More MCP servers
- Context7Pulls current, version-specific library documentation and code examples into an assistant’s context on request.
- Microsoft Learn MCP ServerLets an assistant search and read official Microsoft documentation and code samples, free and with no sign-in.
- DeepWiki MCPLets an assistant read DeepWiki’s generated docs for public GitHub repositories and ask questions about them.
- Brave Search MCP ServerLets an assistant search the web, news, images, videos and local places through the Brave Search API.
- Exa MCP ServerLets an assistant search the web with Exa and read pages as clean Markdown, with optional multi-step research.
- Tavily MCP ServerLets an assistant search the web in real time, extract page content, and map or crawl a website through Tavily.
All MCP serversMCP config generatorGuide: MCP servers in Claude CodeGuide: what is MCP?