Choose a capture scope

The scope controls which nodes enter each document snapshot. It does not remove the configured node, depth, frame, byte, or time limits.

Document

Use document when completeness within accessible boundaries matters more than snapshot size.

await domstamp.capture(page, {
  scope: { mode: 'document' }
});

The walker starts at document.documentElement unless root is a CSS selector.

Viewport

Use viewport for what is currently rendered near the visible window. Matching elements retain their structural ancestors, text, and comments.

await domstamp.capture(page, {
  scope: { mode: 'viewport', margin: 160 }
});

margin expands the intersection boundary in CSS pixels. Elements with display: none or visibility: hidden are excluded by the Playwright adapter.

Focus

Use focus when an agent is blocked around its current control or task context.

await domstamp.capture(page, {
  scope: {
    mode: 'focus',
    siblingCount: 2,
    includeReferences: true
  }
});

Focus scope keeps:

• the deeply focused element, including focus inside open Shadow DOM;
• its accessible subtree and structural ancestors;
• nearby element siblings according to siblingCount;
• labels and supported ARIA references when includeReferences is enabled.

The result is a compact inward snapshot for diagnosis rather than a lossy text summary.

No focus

If the document has no focused element, the selected root becomes the focus anchor.

Root selectors

Every scope accepts root, a CSS selector resolved independently inside each captured document.

scope: { mode: 'document', root: 'main[aria-label="Checkout"]' }

A missing selector is represented as a frame capture error rather than silently falling back to the full document.