Firecrawl
A self-hosted web scraping API converts pages to Markdown, crawls sites, and supports LLM extraction through Ollama or OpenAI-compatible providers.
Firecrawl turns web pages into content that developers can use in AI apps and agents. Its self-hosted API suits teams that want to collect web data on infrastructure they control.
- Page scraping: Retrieve page content as Markdown or HTML, including pages that need JavaScript processing.
- Site crawling and mapping: Collect content across a website or discover its URLs before choosing which pages to scrape.
- Web search: Search the web and retrieve the full page content of results.
- LLM extraction: Extract structured data with Ollama or an OpenAI-compatible provider; this capability requires a connected model backend.
- Agent connections: Connect the self-hosted API to the Firecrawl CLI or a local MCP server for use with AI clients.
Firecrawl is open source under AGPL-3.0 and runs on your machine or server through Docker Compose. Firecrawl Cloud provides a hosted alternative with additional capabilities. Self-hosting keeps the API and its supporting services on your infrastructure, but scraping still contacts target websites. Optional AI, proxy, and parsing providers introduce additional data flows.
The default self-hosted stack doesn't support screenshots or page actions; those require a separate Fire-engine service. Agent, Browser, and interact capabilities require Firecrawl Cloud or additional external services.
