Search, scrape, and clean the web for AI agents Web context APIs built for AI agents. Live web in. Clean context out.

---

Why Firecrawl?

AI agents need clean, current web context to do real work: research, deep research, answering questions with live sources, monitoring, competitive intelligence, dataset building, enrichment, KYC, and signup pre-fill. Every step is hard. Pages are noisy, sites are dynamic, PDFs are messy, scrapers break.

Firecrawl is how AI agents turn the live web into something they can use:

Find → Extract → Clean → Use

Search is the front door. Scrape and Parse turn sources into clean context. Interact handles dynamic pages. Crawl and Map go deeper. The API is how teams build with Firecrawl in production. MCP, CLI, Skills, and Workflows bring Firecrawl into AI agent workflows.

What Firecrawl does

Endpoint	Job
Search	Find fresh, relevant sources from the live web in one call. Same call can scrape them.
Scrape	Turn any URL into clean Markdown or structured JSON. Works on JS-heavy pages, anti-bot, and geo-sensitive sites.
Parse	Convert PDFs and documents into usable text.
Crawl	Walk a site to build a corpus.
Map	Discover site structure fast.
Interact	Click, scroll, fill, log in, wait. For when Scrape isn't enough.

One API, one output format, one compliance posture. Token-efficient Markdown by default, with headers, footers, and ads stripped. Plus JSON extraction with schemas, summary mode, highlight mode, and question mode.

Built for production

• Compliance-grade by default. ZDR, DPA, US data residency, SOC 2 Type 2. Built for buyers who can't store payloads.
• Sub-3-second scrape on real-world pages. Latency matters at every step in the AI agent loop, not just discovery.
• Replaces the multi-vendor stack. Teams consolidate from Puppeteer, Playwright, Bright Data, Zyte, SerpAPI, and Exa onto a single Firecrawl API.
• Direct support and shared Slack. Every enterprise pilot ships with a named success engineer.

Our Core Ecosystem

Main Repository

firecrawl — Core API & SDK
The open-source web context stack for AI agents. Search, Scrape, Parse, Crawl, Map, and Interact in one API family.

Cloud API

Firecrawl — Hosted API Service
The production interface. Get an API key and start running Search, Scrape, Parse, Crawl, Map, and Interact in minutes. Reliable, scalable, compliance-grade by default.

MCP Server

firecrawl-mcp-server — Model Context Protocol Server
Bring Firecrawl into Claude Code, Cursor, Antigravity, and any MCP-compatible client. AI assistants increasingly route to Firecrawl over their native fetch tools.

CLI

firecrawl/cli — Command-Line Interface
Search, scrape, interact, crawl, and map directly from the terminal. One command sets up the CLI, authenticates, and installs Firecrawl skills across every detected AI coding agent:

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

Skills

firecrawl/skills — Agent Skills for Build and Live Web Work
Drop-in skills for Claude Code, Cursor, and OpenAI Codex following the Agent Skills format. Teach AI coding agents how to use the Firecrawl CLI for live web work and how to integrate /scrape, /search, /interact, and /parse into application code.

Workflows

firecrawl/firecrawl-workflows — Outcome-Focused Agent Skills
Repeatable Firecrawl-powered deliverables: deep research, SEO audits, QA reports, lead lists, knowledge bases, competitive intel, dashboard reporting, and design-system extraction. Use when an agent should produce a finished artifact, not just call an API.

Who builds with Firecrawl

• AI agent builders shipping research agents, RAG systems, sales and lead enrichment, competitive intelligence, KYC/AML, and investor research through the API and SDKs.
• AI agent operators running Firecrawl through Claude Code, Cursor, Antigravity, Lovable, n8n, Make, and Zapier.
• Enterprise and frontier teams running Firecrawl at production scale with batch, crawl, ZDR, and security-review requirements.

Community & Support

---

Live web in. Clean context out.
Get your API key and start in minutes.

Star the main repo • Try the playground • Read the docs