POST /scrape

Fetches a page and extracts values from it with CSS selectors.

Try itCopy Anchor Link

POST /scrape
Content-Type: application/json

Request bodyCopy Anchor Link

{
  "target_url": "https://en.wikipedia.org/wiki/Web_scraping",
  "fields": ["h1", "h2"]
}

The body must be a JSON object of at most 6 MB, sent with Content-Type: application/json. Unknown properties are ignored.

ResponseCopy Anchor Link

200 OK

{
  "target_url": "https://en.wikipedia.org/wiki/Web_scraping",
  "final_url": "https://en.wikipedia.org/wiki/Web_scraping",
  "status": 200,
  "data": ["Web scraping", "Contents"],
  "warnings": [],
  "meta": {
    "id": "5f0c2d8e-3b1a-4c6e-9d2f-7a8b9c0d1e2f",
    "started_at": "2026-10-24T09:12:03.418Z",
    "duration_ms": 412,
    "engine_ms": 388,
    "cached": false,
    "attempts": 1,
    "strategy": 0,
    "bytes": 184213,
    "proxy": null,
    "fingerprint": null,
    "browser": null
  }
}

What the page holds comes first (data, then what the options add); how the request went is grouped in meta, last. With several pages (steps, /batch, /crawl), each page has its own meta (without id and started_at), and the response's meta describes the whole request.

metaCopy Anchor Link

Failed requests return an error object instead of data. See Errors.

FieldsCopy Anchor Link

The simplest field is a CSS selector string. It returns the text of the first matching element, or null if nothing matched:

"fields": ["h1", "h2"]

When you need options, write the field as an object. "h1" and { "selector": "h1" } are the same field.

Strings and objects can be mixed in the same fields.

Get every matchCopy Anchor Link

{
  "data": [
    "Web scraping",
    ["Contents", "History", "Techniques", "…"]
  ]
}

With "multiple": true the value is always an array, [] if nothing matched, even when the page has a single match.

Name your valuesCopy Anchor Link

With more than a couple of fields, positions get hard to follow. Pass fields as an object instead: each key is a name of your choice, and data comes back as an object with the same names.

{
  "data": {
    "title": "Web scraping",
    "canonical": "https://en.wikipedia.org/wiki/Web_scraping",
    "sections": ["Contents", "History", "Techniques", "…"]
  }
}

Names start with a letter or _ and contain only letters, digits, _ or -, 64 characters max.

Comma-separated selectorsCopy Anchor Link

A selector such as "h1, h2" always returns an array, with the selectors in the order you wrote them, not in the order they appear in the page:

• without multiple: the first match of each selector, e.g. ["Web scraping", "Contents"];
• with "multiple": true: every match of the first selector, then every match of the next, e.g. ["Web scraping", "Contents", "History", …].

A selector that matches nothing is left out of the array.

Value typesCopy Anchor Link

The type of each value depends only on your request, never on the page, so your code always receives the type it expects:

Values are cleaned: whitespace collapsed and trimmed, control characters removed, 10,000 characters max per value, 1,000 values max per array.

OptionsCopy Anchor Link

"options": {
  "method": "POST",
  "headers": {
    "Content-Type": "application/x-www-form-urlencoded"
  },
  "body": "q=shoes",
  "retry": 3,
  "backoff": { "initial_delay": 1000 }
}

Turn a feature offCopy Anchor Link

Set an option to false to turn it off explicitly:

In a fallback strategy, false also stops inheriting from the main request: "proxy": false, "fingerprint": false, "browser": false, "retry": false, and "headers": false, "cookies": false or "body": false to send none of the main ones.

timeout and max_redirects can't be false: every request has a timeout, and "follow_redirects": false is the way to stop following redirects.