Firecrawl: an open source project overview

Core facts

Read by a script from the official API, not rewritten by a model.

Repository: https://github.com/firecrawl/firecrawl
License: AGPL-3.0
Languages: TypeScript
Maintainer: firecrawl
Homepage: https://firecrawl.dev
Verified: 2026-10-31

What it is

Firecrawl is open source under the AGPL-3.0 license. This project is primarily licensed under the GNU Affero General Public License v3.0 (AGPL-3.0). The SDKs and some UI components are licensed under the MIT License. The cloud version includes additional features.

Main features

Search searches the web and returns full page content from results.
Scrape converts a URL to markdown, HTML, screenshots, or structured JSON.
Interact scrapes a page, then interacts with it using AI prompts or code.
Crawl crawls a website and gets page content.
Map discovers URLs on a website.
The SDKs handle polling automatically.
In Agent, the effort setting changes the reasoning budget, not the model.
A Crawl request returns a job ID.
Agent is the evolution of the /extract endpoint.
Every Agent run executes on the spark-2 model.
Map with search can find specific URLs within a site.
In Agent, the low effort level is for simple lookups on one site.
The medium effort level suits multi-step tasks on a few pages.
The high effort level suits deep research, complex navigation, and critical data.

Getting started

Signing up gives you an API key.
You can try the playground to test it out.
The skill gives an agent access to real-time web data with one command.
Restart the agent after installing.
An AI agent can fetch this skill to sign up its user, get an API key, and start building with Firecrawl.
For MCP, the README points to the separate firecrawl-mcp-server repository.
For self-hosting, the README points to a separate Self-Hosting Guide.

From the README · bash

npx -y firecrawl-cli@latest init --all --browser

From the README · json

{
  "mcpServers": {
    "firecrawl-mcp": {
      "command": "npx",
      "args": ["-y", "firecrawl-mcp"],
      "env": {
        "FIRECRAWL_API_KEY": "fc-YOUR_API_KEY"
      }
    }
  }
}

From the README · bash

curl -s https://firecrawl.dev/agent-onboarding/SKILL.md

Who it is for

Parse turns PDFs, DOCX, and other documents into clean Markdown or structured JSON. Actions click, scroll, write, wait, and press before extracting content. With the Agent endpoint, you describe what you need and the AI agent searches, navigates, and retrieves it.

Agent offers automated data gathering, where you just describe what you need. The agent needs no URLs. Monitor notifies you when content on a website changes.

By default, Firecrawl respects robots.txt directives. Respecting websites' policies when scraping is the sole responsibility of end users. Users are advised to adhere to applicable privacy policies and terms of use.

Related tools listed on this site.