Capture Tasks

A capture task is one page of a capture batch: what was asked, what the page answered, and the screenshot.

Read a capture task

endpoint

GET /api/v1/projects/{project_id}/capture-tasks/{capture_task_id}

One capture_task. Minimum key role: Participant. Its id is in the batch's tasks.

The `capture_task` object

capture_task

{
  "object": "capture_task",
  "id": "00000000-0000-0000-0000-000000000005",
  "capture_batch_id": "00000000-0000-0000-0000-000000000002",
  "page": {
    "id": "00000000-0000-0000-0000-000000000007",
    "path": "/pricing",
    "title": "Pricing"
  },
  "status": "DONE",
  "error": null,
  "input": {
    "url": "https://example.com/pricing",
    "viewport": {
      "width": 1280,
      "height": 800
    },
    "max_capture_height": 20000,
    "mask_selectors": [
      ".ad"
    ],
    "hide_selectors": [
      "#cookie-banner"
    ],
    "unfocus_before_capture": true,
    "freeze_animations": true,
    "settle_delay_ms": 0
  },
  "custom_headers_used": [
    {
      "name": "x-preview-token",
      "sent_to_third_parties": false
    }
  ],
  "response": {
    "http_status": 200,
    "page_title": "Pricing – Example"
  },
  "capture": {
    "document": {
      "width": 1280,
      "height": 3120
    },
    "captured": {
      "width": 1280,
      "height": 3120
    },
    "cropped_width": 0,
    "truncated_height": 0,
    "fonts_ready": "loaded"
  },
  "dom_prep": {
    "hidden": [
      {
        "selector": "#cookie-banner",
        "count": 1
      }
    ],
    "painted": [
      {
        "selector": ".ad",
        "count": 0
      }
    ],
    "invalid": [],
    "unfocused": true,
    "animations": {
      "finished": 3,
      "cancelled": 1
    }
  },
  "crashed_frames": [],
  "renderer": {
    "chrome": "155.0.8059.79",
    "built_at": "2026-10-26T14:03:12.511Z",
    "fonts": {
      "sans-serif": [
        "Liberation Sans"
      ]
    }
  },
  "timing": {
    "load_ms": 1840
  },
  "screenshot": {
    "url": "https://…",
    "expires_at": "2026-11-06T12:00:00.000Z",
    "bytes": 482133,
    "sha256": "9f86d081884c7d659a2feaa0c55ad015a3bf4f1b2b0b822cd15d6c15b0f00a08"
  },
  "thumbnail": {
    "url": "https://…",
    "expires_at": "2026-11-06T12:00:00.000Z",
    "bytes": 21004
  },
  "created_at": "2026-11-06T10:00:00.31+00:00",
  "updated_at": "2026-11-06T10:00:09.7+00:00"
}

Identity and outcome

Field	Type	Value
`capture_batch_id`	uuid string	Its batch.
`page`	object \| null	The page: `id`, `path`, `title` today. `null` once the page is deleted.
`status`	string	`PENDING`, `RUNNING`, `DONE` or `ERROR`.
`error`	object \| null	When the capture failed: `code` and `message` — see when a page capture fails. `null` otherwise.

A capture that failed because of the page — an error status, a timeout, a redirect away — still has what it reached: response, capture, screenshot and the rest may be filled in.

What was asked

input is the capture as it was set up when it ran — the page's settings may have changed since. See capture config.

Field	Type	Value
`input.url`	string \| null	The address captured.
`input.viewport`	object \| null	`width` and `height` in pixels: the browser's window. The width is held to the plan's maximum.
`input.max_capture_height`	number \| null	The tallest capture the plan allows, in pixels; `null` when the plan sets none.
`input.mask_selectors`	string[]	CSS selectors painted over in a solid colour, layout kept.
`input.hide_selectors`	string[]	CSS selectors hidden, layout collapsed.
`input.unfocus_before_capture`	boolean \| null	Whether focus was taken off the page before the shot, so no field shows a caret or ring.
`input.freeze_animations`	boolean \| null	Whether animations were brought to their end, endless ones removed, before the shot.
`input.settle_delay_ms`	number \| null	A wait after the page loaded, in milliseconds, from 0 to 5000.
`custom_headers_used`	array	The environment's custom HTTP headers sent: their `name` only, never their value, and `sent_to_third_parties`, whether they also went to other sites the page loads from.

What the page answered

Field	Type	Value
`response.http_status`	number \| null	The page's HTTP status.
`response.page_title`	string \| null	The page's `<title>` when it was captured.
`capture`	object \| null	The sizes, in pixels. `null` when nothing was captured.
`capture.document`	object	`width` and `height` of the page itself, measured just before the shot.
`capture.captured`	object	`width` and `height` of the screenshot.
`capture.cropped_width`	number	Pixels of horizontal overflow cut off. `0`: nothing lost.
`capture.truncated_height`	number	Pixels of page height cut off at the plan's limit. `0`: nothing lost.
`capture.fonts_ready`	string \| null	`loaded` when the web fonts were ready before the shot; `timeout` when they were not, and fallback fonts may show.
`dom_prep`	object \| null	What preparing the page did. `null` when it did not run.
`dom_prep.hidden`	array	Per hide selector: `selector`, and `count`, how many elements it matched. `0` is normal on a page without them.
`dom_prep.painted`	array	The same for the mask selectors.
`dom_prep.invalid`	string[]	Selectors the browser rejected as invalid CSS. They are skipped, the capture goes on.
`dom_prep.unfocused`	boolean \| null	Whether focus was taken off the page.
`dom_prep.animations`	object \| null	`finished` and `cancelled`: animations brought to their end, and endless ones removed. `null` when animations were left running.
`crashed_frames`	string[]	Addresses of frames from other sites that crashed during the capture. The page is captured all the same, with a grey box where each was — see when an embedded frame crashes.
`renderer`	object \| null	The browser that drew the page.
`renderer.chrome`	string \| null	Its Chrome version.
`renderer.built_at`	string \| null	When that browser image was built.
`renderer.fonts`	object \| null	Per font family the browser was tested with, the fonts it actually drew with. Two captures drawn with different fonts can differ for that alone.
`timing.load_ms`	number \| null	How long the page took to load, until its network went quiet, in milliseconds.

Images

Field	Type	Value
`screenshot`	object \| null	The full-page PNG. `null` when there is none.
`screenshot.url`	string \| null	A signed address to download it from, valid until `expires_at`.
`screenshot.expires_at`	timestamp \| null	When `url` stops working: up to two hours after the call. Trust this date, not a fixed delay.
`screenshot.bytes`	number \| null	Its size, approximately.
`screenshot.sha256`	string \| null	The SHA-256 of the PNG, in hex. Two captures with the same hash are the same image, byte for byte.
`thumbnail`	object \| null	A small preview of the screenshot: `url`, `expires_at`, `bytes`, as above.

Download an image while its url is valid, or read the task again for a fresh one: do not store the address. url is null in the rare case the image could not be signed at that moment; read the task again.