Skip to main content
Scrape visits a URL and returns the outputs you enable in formats. Combine formats in one request and check each output’s success before using its data.

Make a request

Set up your API key and SDK in the Quickstart. This request returns the page’s main content as Markdown.

Read the response

Every output has requested, success, and data. A successful output has success: true. Failed outputs have success: false and data: null; outputs you did not request have success: null. One failed output can coexist with successful ones in an HTTP 200 response. Abridged response:
isPartial flags incomplete results. metadata describes the page; key_metadata reports request usage. Save request_id when investigating failures.

Formats

Shape the page visit

Use sharedParams for browser actions, waiting and rendering, content filtering, location and headers, or PDF parsing. Use maxAgeMs for freshness, timeoutOpts for deadlines, zdr for zero data retention, and tags for usage tracking. The reference lists the complete request and response.

Choose another API

Use Map to find a site’s URLs, Crawl to read several pages, or Batches to process a URL list asynchronously. Search finds pages from a query; Answers researches a question. Parse accepts uploaded files.

Move from the per-format endpoints

The older routes remain available. New integrations can request several formats through POST /web/scrape: The option names and response shapes differ. Use the format guides when migrating.