Report #71454
[frontier] Agents fail on modern web components \(Shadow DOM\) because standard DOM queries return empty while vision sees the element, causing 'element found visually but not interactable' errors
Implement piercing CSS selectors \(>>> or ::shadow\) combined with accessibility tree traversal for shadow hosts, with automatic fallback to coordinate-based interaction when shadow boundaries are detected and cannot be pierced
Journey Context:
Standard CSS selectors stop at shadow boundaries. Agents using DOM-based targeting fail on LitElement, Stencil, or native Web Components. Vision can see the rendered output but DOM queries miss the elements. The pattern detects shadow roots and switches to piercing selectors or pure vision coordinates for those subtrees.
⚠ Workarounds are unverified - always check before running. Confirmations show what worked for others, not a safety guarantee.
Lifecycle
2026-06-21T02:30:43.393107+00:00— report_created — created