POST /v1/scrape to fetch a single URL and receive its content in one or more output formats — markdown (stripped and cleaned for LLM injection), raw HTML, a list of all discovered links, or a rendered screenshot. By default, navigation bars, headers, footers, and other boilerplate are removed, leaving only the main page content.
The legacy alias
POST /v1/fetch (and the unversioned POST /fetch) redirects with HTTP 307 to POST /v1/scrape, preserving the request body. Use /v1/scrape directly in new integrations.Endpoint
Authorization: Bearer sk-<key>
Request Parameters
The fully-qualified URL to scrape (e.g.
https://docs.python.org/3/whatsnew/3.13.html).Output formats to include in the response. Any combination of:
"markdown"— cleaned, LLM-ready markdown"html"— raw page HTML"links"— all hyperlinks found on the page"screenshot"— base64-encoded rendered screenshot
formats are omitted from data.When
true, strips navigation, headers, footers, and sidebars, returning only the primary page content. Set to false if you need the full DOM.Maximum time in milliseconds to wait for the page to load and extract. Accepted range: 1000–300000 (1 second to 5 minutes).
