scrape-website is useful when an agent needs to fetch the same public page again after its content changes. The listing now makes that recurring job explicit: send the URL, choose text, HTML, or markdown, and save the returned content and metadata for comparison on the next run.

The response includes the final URL after redirects, status, title, meta description, canonical URL, OpenGraph and Twitter card metadata, headings, body content, and optional links. Static pages and server-rendered sites fit this route; use website-screenshot when the job needs a rendered JS-heavy page.