threejs-game-image-optimization
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data in the form of image files and project metadata.
- Ingestion points: The skill reads image files (PNG, JPEG, WebP, KTX2), 3D model files (glTF/GLB), and project manifests from the local filesystem.
- Boundary markers: The instructions do not explicitly define boundary markers or provide warnings to ignore potential instructions embedded in image metadata or model tags.
- Capability inventory: The skill is capable of performing filesystem writes (replacing original assets) and executing external optimization tools with configurable parameters via the agent's environment.
- Sanitization: The skill performs structural validation (decoding checks) and quality verification (pixel comparisons and rendering checks) which serves as a functional sanitization layer against malformed data.
Audit Metadata