Choose desktop, mobile, viewport, or bounded full-page capture.
Screenshot API for AI agents
URL in. Screenshot out. →
Give AI agents eyes on the web.
Our screenshot API turns any public webpage into a clean PNG or JPEG, with its source and render details attached. Agents can inspect a page, verify what changed, compare states, and hand visual evidence to a human.
Development preview · production screenshots and signup are not live yet
SCREENSHOT RESULT
A webpage your agent can see.
Clean pixels plus the context needed to use them.
Screenshot ready
1440 × 900 · PNG
POST
/v1/screenshots
202 queued
Retrieve a clean PNG or JPEG from a private, expiring URL.
Keep the final URL, dimensions, viewport, status, and timing.
A vision tool for AI agents
Let an agent look at a webpage—not just read its HTML.
HTML and extracted text miss layout, visual hierarchy, charts, broken states, overlays, and what a user actually sees. A screenshot gives multimodal agents the rendered page as visual input, with metadata that keeps the observation traceable.
- See the pageCapture responsive layouts, images, charts, modals, and rendered UI.
- Know the sourcePair every image with its requested URL, final URL, viewport, and timing.
- Use it as a toolGive an agent one strict request schema and a durable screenshot job.
RequestResponse
POST /v1/screenshots
{
"url": "https://example.com",
"preset": "desktop",
"full_page": true,
"format": "png"
}
202 Accepted · shot_01K...
Website screenshots, under API control
Capture the page your agent actually needs to see.
Choose the screen size, page length, image format, wait behavior, and limits instead of running your own browser fleet.
Desktop and mobile screenshots
Render a webpage at explicit viewport sizes so an agent can inspect the right responsive layout.
Viewport or full-page capture
Capture what is visible above the fold or scroll through a bounded page to create one complete image.
Private screenshot artifacts
Store results privately and give the calling workflow short-lived access instead of publishing permanent image URLs.
From URL to screenshot
Three steps an agent can call reliably.
- 1Send the webpage URLAsk for desktop, mobile, viewport, or full-page output.
- 2We capture the screenshotAn isolated browser opens the page and creates the image.
- 3Your agent uses the resultRetrieve the private image and its page context from one job.
Why agents need screenshots
Give AI workflows eyes on the rendered web.
Use website screenshots whenever pixels, layout, responsive behavior, or visible state matter—not only the page source.
See charts, layouts, products, and visual context.
Add a screenshot to extracted text so a multimodal agent can understand what the webpage looks like as well as what it says.
Explore the agent workflow →Check what a user would actually see.
Capture a release, generated page, form state, or mobile layout and attach the image to an automated review.
Read the full-page guide →Keep visual evidence when a page changes.
Save repeatable screenshots for debugging, audits, content checks, change investigations, and human handoff.
Learn the API workflow →
Clear launch boundaries
Know what the API will—and will not—do.
The MVP focuses on public webpages, bounded rendering, isolated browser workers, and private artifacts. Arbitrary JavaScript, credential-bearing URLs, persistent browser sessions, customer proxies, and CAPTCHA bypass are outside the first release.
Read every product answer →Is Screenshot API live?
The website and API contract are public previews. Customer signup and production screenshot rendering remain disabled.
Will artifacts be private?
The planned design uses encrypted private object storage and short-lived signed download URLs.
Can it capture full pages?
Yes, within documented page-height, execution-time, and artifact-size limits.
Now in development
Read the first contract →