webgl-dom-synced-planes

Pass

Audited by Gen Agent Trust Hub on Aug 30, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill implementation reads asset URLs directly from DOM element attributes, which constitutes an ingestion point for untrusted data if the HTML context is externally controlled.
  • Ingestion points: In SKILL.md, the functions createPlane and addPlane retrieve resource URLs from src, data-src, and data-video attributes of target DOM elements.
  • Boundary markers: The instructions do not define boundary markers or explicit isolation to prevent the interpretation of data as instructions.
  • Capability inventory: The generated code utilizes THREE.TextureLoader and HTMLVideoElement to perform network requests to the extracted URLs.
  • Sanitization: The provided code snippets do not implement URL validation, protocol checking, or domain whitelisting to prevent loading content from potentially malicious external sources.
  • [EXTERNAL_DOWNLOADS]: The skill facilitates the download of external texture and video assets from URLs provided at runtime. This allows the application to communicate with arbitrary external servers, though this is typical for WebGL asset loading.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 30, 2026, 05:49 PM
Security Audit — agent-trust-hub — webgl-dom-synced-planes