# Firecrawl Firecrawl turns websites into **LLM-ready data**. Give it a URL and it returns clean Markdown or structured JSON — JavaScript rendering, proxies, and anti-bot handling included — so agents and RAG pipelines read the web without you writing scraper code per site. ## The API surface - **/scrape** — one URL → clean Markdown/structured output - **/crawl** — whole site traversal, every page converted - **/map** — fast URL discovery for a site - **/extract** — schema-driven structured extraction across pages (describe what you want, get typed JSON) Open-source core with a hosted API as the business; SDKs for Python and Node, plus integrations into [[LlamaIndex]], LangChain, and agent frameworks. ## Why it matters Web access is the most common tool an [[AI Agents|agent]] needs after file access, and raw HTML wastes context on markup. "URL in, Markdown out" became the standard contract for that layer. Firecrawl's [[pdf-inspector]] applies the company's cost obsession to PDFs: classify first, OCR only what needs it. ## References - [Firecrawl Website](https://www.firecrawl.dev) - [Firecrawl on GitHub](https://github.com/firecrawl/firecrawl) ## Related - [[pdf-inspector]] — their PDF triage library - [[LlamaIndex]] — common downstream consumer - [[Retrieval-Augmented Generation (RAG)]] - [[AI Agents]]