SnapScout: Website Screenshot API for AI Agents (Image + Text)

Let your agent see any web page, and read it too. One call per URL returns a public screenshot URL plus the page's status, title, meta tags and clean visible text, with cookie banners already clicked away.

Why SnapScout

• Clean shots. Consent banners (OneTrust, Cookiebot, Didomi, Quantcast, TrustArc, Usercentrics, CookieYes and plain "Accept all" buttons in 10+ languages) are dismissed before capture, so you see the page, not the overlay.
• Image and text in one record. Agents can read the page without a vision model, and QA scripts can diff status, title and text without comparing images.
• $0.003 per screenshot, failures free. Unreachable pages, timeouts and missing selectors aren't charged.

Example record (trimmed):

{ "url": "https://example.com/", "ok": true, "httpStatus": 200, "title": "Example Domain",
  "screenshotUrl": "https://api.apify.com/v2/key-value-stores/…/records/001-example-com.png",
  "text": "Example Domain\nThis domain is for use in documentation examples…", "cookieBannerDismissed": false }

What you get

Give it URLs. For each one you get one record:

• a public screenshot URL (PNG or JPEG; first screen, full page or one element)
• the HTTP status and final URL after redirects
• title, meta description, first H1, language, canonical URL and og:image
• the page's visible text, cleaned and cut to the length you choose, so an agent can read the page without a vision model
• whether a cookie banner was clicked away, plus a warning if the page never finished loading

Who it's for

• AI agents and MCP clients that need to see a page ("what does competitor.com's pricing page look like?", "did the deploy break the homepage?") or quote it with evidence.
• QA and monitoring: screenshot a list of pages on every deploy; the status, title and text let you diff without comparing images.
• Sales and research: one image plus summary facts per prospect website.
• Content and SEO: capture title, description, H1 and og:image next to how the page actually renders.

Input

{
  "urls": ["https://apify.com", "example.com", "https://stripe.com/pricing"],
  "device": "desktop",
  "fullPage": false,
  "format": "png",
  "includeText": true,
  "maxTextChars": 4000
}

Field	Default	What it does
urls	required	Full URLs or bare domains. One record each; duplicates skipped.
device	desktop	desktop 1366x768, laptop 1920x1080, tablet 820x1180 @2x, mobile 390x844 @2x.
fullPage	false	Whole scrollable page (scrolls first so lazy images load), capped at maxFullPageHeight (15,000 px).
selector	none	Capture one element, e.g. #pricing.
format / jpegQuality	png / 80	JPEG is 3-10x smaller, good for vision models.
includeText / maxTextChars	true / 4000	Visible text of main/article (else body).
dismissCookieBanners	true	Click "accept", then hide leftover overlays.
waitUntil	load	domcontentloaded, load or networkidle, best effort up to 15 s. Pages that never settle are still captured, with a warning.
delaySecs	1	Extra wait after loading.
includeImageBase64	false	Also embed the image bytes in the record.
viewportWidth / viewportHeight	device	Custom viewport.
maxConcurrency	2	Pages captured in parallel (1-8).

Output

One dataset item per URL:

{
  "url": "https://example.com/",
  "ok": true,
  "httpStatus": 200,
  "finalUrl": "https://example.com/",
  "screenshotUrl": "https://api.apify.com/v2/key-value-stores/…/records/001-example-com.png",
  "title": "Example Domain",
  "description": null,
  "h1": "Example Domain",
  "lang": "en",
  "text": "Example Domain\nThis domain is for use in documentation examples without needing permission. …",
  "textChars": 127,
  "textTruncated": false,
  "cookieBannerDismissed": false,
  "device": "desktop",
  "viewport": { "width": 1366, "height": 768 },
  "fullPage": false,
  "format": "png",
  "bytes": 19675,
  "pageHeight": 768,
  "linkCount": 1,
  "loadMs": 1840,
  "capturedAt": "2026-09-23T15:20:00.000Z"
}

Failed pages come back with "ok": false and an error (for example net::ERR_NAME_NOT_RESOLVED or selector not found: #pricing) and are not charged. Pages that return 404 or 500 are still captured, with their status, because that's often what you want to see.

The images are also in the run's key-value store (001-example-com.png, 002-…), in input order.

Pricing

$0.003 per screenshot captured, with no start fee and no subscription. Failed pages are free. 1,000 screenshots cost $3.

Use it from an AI agent

• Apify MCP server (https://mcp.apify.com): add the Actor cybermax/website-screenshot and call it with {"urls": [...]}. Agents without an Apify account can pay per call through Apify's agentic payments.
• API: POST https://api.apify.com/v2/acts/cybermax~website-screenshot/run-sync-get-dataset-items?token=<TOKEN> with the input JSON returns the records directly.

When not to use it

• You only need the text or HTML. A plain fetcher is cheaper and faster.
• Pages behind a login, CAPTCHA or strong bot protection. This Actor uses a normal headless Chrome with no proxy, so some sites (large retailers, social networks) may show a block page. The record still tells you what was shown.
• Pixel-perfect visual regression. Fonts, ads and A/B tests change between captures.

Limits and notes

• Chrome runs headless on Apify, with no proxy by default. Geo-specific content reflects Apify's data-centre location (US/EU).
• Full-page captures are capped at 15,000 px tall by default (configurable up to 30,000); longer pages are cut, with a warning.
• Records are returned in completion order; screenshotKey numbers follow input order.

Built by CyberMax. More tools: apify.com/cybermax.