threejs-loaders
Pass
Audited by Gen Agent Trust Hub on Nov 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the loading and processing of external 3D models and textures, creating a surface for indirect prompt injection or resource exhaustion via maliciously crafted assets.
- Ingestion points: Asset loading occurs through
loader.load()calls for various formats (GLTF, OBJ, textures) inreferences/detailed-guide.md. - Boundary markers: The guide advises implementing allowlists for resource origins and bounding file/dimension budgets to mitigate risks.
- Capability inventory: The skill enables network requests for sub-resources and complex scene graph traversal.
- Sanitization: Includes basic validation logic such as checking model bounds using
Number.isFinite(maxDim). - [EXTERNAL_DOWNLOADS]: The skill configures Three.js loaders to fetch necessary decoders and transcoders from external CDNs.
- Evidence: Fetches Draco decoders from Google's
gstatic.comand KTX2 transcoders fromjsdelivr.net, which are established and well-known services for these assets.
Audit Metadata