Skip to main content
POST /v1/scrape fetches one public page and returns the formats you request. The endpoint requires a key with the scrape scope. See API keys to create or rotate one.

​
Request a page

Provide a public URL and choose the formats your application needs. If you omit formats, the API returns Markdown. Both a format array and an object of enabled formats are supported.
Request fields accept camelCase and snake_case. See the API reference for all parameters, generated from the implemented OpenAPI contract. Configure content outputs and browser settings to match your source. Each ordinary call performs a fresh fetch. Cache controls such as maxAge, minAge, and storeInCache are not accepted by the current API. See quote and credits for billing behavior.

​
Response

The response wraps the result in data and includes a top-level request_id. data.metadata reports source_url, final_url, status_code, and credits_used. It can also report the page title, description, author, language, publication date, and capture identifiers. Requested outputs are returned in data; outputs you did not request are null. html is cleaned content, and raw_html contains the fetched HTML. A successful scrape produces every requested output.

​
Screenshot options

By default, screenshots are viewport captures encoded as JPEG with quality 80 and returned as inline data URLs. Set fullPage: true, or use screenshot@fullPage in a format array, to capture the full page. Do not combine screenshot@fullPage with an explicit fullPage: false. For PNG, omit quality. For a public image link, set delivery: "url". URL delivery waits for image storage, while base64 delivery does not. Public image URLs have no signature expiry and remain usable while the file is stored. Anyone with the URL can access the image. Read the screenshot guide for encoding, capture restrictions, and delivery tradeoffs.

​
Retries

Send an Idempotency-Key header to avoid repeating work or charges when retrying the same scrape or map request. Reuse the key with identical inputs. A retry replays the saved response when available; otherwise it returns 409 with code IDEMPOTENCY_KEY_ALREADY_ACCEPTED, without executing again. Saved responses may be delayed or unavailable. That conflict does not mean the original request failed. Reusing a key with different inputs returns 422 with code IDEMPOTENCY_KEY_CONFLICT. Use a new key only for an intentional new request. See retries and idempotency for a complete retry policy. Credits are reserved before execution and charged when completed. Blocks, timeouts, and server errors release the reservation. A returned 404 page is billable. Activity shows the failure reason, credits charged, and credits released.