Firecrawl
Firecrawl is a web data API: it searches the web, scrapes and crawls pages, and returns them as clean Markdown or structured JSON for an app or AI agent to use.
It suits founders building an AI product that reads the live web, such as a research agent, a support bot or a price monitor, who would rather not run scrapers and proxies themselves.
The free plan gives 1,000 credits a month, about 1,000 pages; paid plans start at $19 a month for 5,000, as checked on 29 October 2026.
Reviewed by the IndieHacked team
- What it does
- An API that turns web pages into clean Markdown or JSON for your app or AI agent: search, scrape, crawl, and click through the pages that need it.
- Our take
- Founders building an AI product that reads the live web. Point it at a URL or a search and get Markdown or JSON back, or let Claude Code, Cursor or Codex call it through Firecrawl's MCP server while they build.
- Pricing
- FreemiumLow price1,000 free credits every month, with no card. Hobby is $19 a month for 5,000 credits, Standard $99 for 100,000, Growth $399 for 500,000 and Scale $749 for 1,000,000; billed yearly, Hobby is $16 a month, as checked on 29 October 2026. A plain scrape costs 1 credit a page and JSON extraction adds 4. Unused credits roll over on Scale only, and payments are not refunded.
The IndieHacked breakdown
What it does
Firecrawl is an API for web data. Scrape turns one page into clean Markdown, HTML, a screenshot or JSON shaped by your own schema; Crawl follows a whole site, Map lists its URLs, and Search runs a web search and scrapes the results. Interact clicks, fills forms and signs in by prompt or Playwright code, and Agent researches a question across sites on its own.
You call it from official SDKs for Python, Node, Go, Rust and more, from a command-line tool, or through its MCP server, which Claude Code, Cursor, Codex and ChatGPT can use; n8n, Zapier and Make reach it without code. The core is open source under the AGPL-3.0 licence, but anti-bot handling, Interact and Agent run on Firecrawl's cloud only.
Who it suits
Founders building an AI product that reads the live web: a research agent, a support bot that answers from your docs, a tool that watches competitors' prices. Firecrawl handles rendering, proxies and the clean-up, so a page reaches your model as Markdown without the menus and scripts around it.
It also fits a founder's own work with an agent. Through the MCP server or the command line, Claude Code or Codex can read documentation or a competitor's site mid-task, and the free plan's 1,000 credits a month cover about 1,000 plain pages. Monitor checks pages on a schedule and tells you by webhook, email or Slack when they change.
Not the right fit for
Developers who want the crawler in their own code, under their own control. Firecrawl's own comparison says Crawl4AI, a free Python library, suits "Python-native teams who need maximum control over the crawler." Firecrawl can be self-hosted too, but its docs recommend the cloud unless you are ready to run it.
Founders who need ready-made scrapers for Instagram, TikTok or Google Maps, or a bill that stays flat. Firecrawl's comparison points to Apify for platform scrapers, and the cost moves with what you ask for: JSON extraction makes a page 5 credits instead of 1, most Agent runs take a few hundred, and unused credits below Scale do not roll over.
Report an error. Spotted a factual or language mistake? .