Firecrawl

The scrape endpoint returns the main content of a page, with navigation and ads removed, in Markdown, HTML, links, screenshots or JSON that follows your schema. Structured data extraction works from that same call, so results land in fields your database can take. PDFs and DOCX files go through the same route, so a mixed pile of URLs and documents can land in one dataset. Output is cached, which cuts repeat costs on pages that rarely change.

Search sends a query to the live web and returns the full text of each result, not just titles and snippets. You can filter by domain, date or country and ask for highlights or a summary. One call gives an agent enough context to answer without a second scraping round.

Crawl starts at a URL and follows links within the rules you set: depth, path patterns, page limits and sitemap use. Jobs run in the background and report progress, so a 5,000 page site doesn't block your app. You get one Markdown or JSON record per page.

Map returns the URL structure of a domain in seconds, which is useful when you don't know how a site is organized. Filter the list after the call and send only the pages you need to crawl. It's cheaper than crawling everything to find out what exists.

Interact opens a real browser session and runs actions such as clicking, typing, scrolling and logging in. That covers pages that only load content after a click or a sign-in. Sessions are meant for short tasks, not for keeping a browser open all day.

Monitor checks a page, a crawl or a search on a schedule you set. When content changes or a new result appears, Firecrawl sends a webhook, an email or a Slack message. This replaces a stack of cron jobs and diff scripts for price tracking, policy pages and release notes.