Skip to main content
Scrinly captures pixel-accurate webpages and turns them into structured visual context agents and automations can inspect. Start with a JPEG or PNG, add deterministic regions when a model needs smaller inputs, generate a screenshot-grounded Visual Style Guide, compare captures, or monitor a page over time.

Capture a screenshot

Return image bytes or a direct B2, S3, or R2 URL.

Split it into regions

Get ordered crops, bitmap bounds, hashes, and coverage.

Compare two captures

Produce perceptual or pixel changes with optional visualization.

Monitor a page

Run fresh captures against rolling or fixed baselines.

Connect an AI agent

Connect Codex, Claude Code, and other MCP clients with scoped OAuth—no API key copied into the client.

Install the Scrinly skill

Teach compatible agents the safe capture, comparison, polling, and spend workflows.

​
Public API surface

The public API is intentionally focused: screenshots, screenshot batches, visual diffs, monitors, asynchronous status, and usage. Structured page extraction, site-wide traversal, generic scheduling, and recovery tooling are internal capabilities and are not customer APIs. Every customer route uses a Scrinly bearer key. Images stored by Scrinly or in your own provider are returned through direct provider URLs; Scrinly does not proxy those assets.
Capture only publicly accessible pages or pages you are authorized to access. Headers, cookies, interactions, and customer-storage credentials belong in HTTPS headers or JSON bodies, never in URLs or client-side code.