html-in-canvas
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill identifies and warns against actions that trigger browser security errors, such as canvas tainting (SecurityError) when using getImageData on captured elements.
- [SAFE]: The documentation mandates using locally bundled vendor scripts and forbids external CDNs to minimize supply chain risks.
- [INDIRECT_PROMPT_INJECTION]: The skill provides an interface for rendering HTML strings, creating a vulnerability surface for injected instructions. 1. Ingestion points: The 'html' parameter in the CanvasUI.create configuration. 2. Boundary markers: Absent; the code does not use delimiters to wrap the injected HTML. 3. Capability inventory: Uses the drawElementImage API and event forwarding for input/textarea/select elements. 4. Sanitization: No explicit sanitization or filtering is implemented in the boilerplate skeleton.
Audit Metadata