Long Page Screenshot

Problem

When reviewing a web page, UI state, or a paired device in chat, a text snapshot can describe structure but cannot show layout, colors, charts, or where an error appears. Engineers or users asking to “capture it” or “show me this page” need a way to deliver screenshots as inline chat media, covering full browser pages, selected elements, and device screens.

How it works

The skill turns screenshots into image replies instead of returning file paths only.

• Browser capture: use browser open to load the target URL, then call screenshot after the page is ready.
• Full page and regions: fullPage captures the whole page; selector captures an area using a CSS selector; maxWidth can limit output width.
• Device screens: check nodes with nodes status, then use nodes camera_snap or nodes screen_record to get the result.
• Delivery: write MEDIA: on its own line in the reply, and the system inlines the image in chat.

Boundaries

Confirm the page has fully loaded before capturing; snapshot can be used as a check. Remind users before capturing pages with passwords or personal data. If capture fails, fall back to snapshot for a text page snapshot. It is useful for quick UI checks, sharing page fragments, and inspecting device state, but it is not a full screen-recording analysis tool or a visual proof system for automation.