blender-render-compositing
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data in the form of Blender scene files, textures, and 3D assets.
- Ingestion points: Operations involve loading 'approved local assets', specific target objects, and texture dependencies into the Blender environment as defined in SKILL.md and examples/request.md.
- Boundary markers: The skill explicitly requires the use of an 'approved output root' and 'approved' directories to scope and isolate the impact of file system operations.
- Capability inventory: The agent has the capability to perform hardware-probed rendering (Cycles/Eevee), bake materials, pack dependencies into archives, and write files to the local filesystem.
- Sanitization: While the direct sanitization of complex binary 3D data is not explicitly performed by the skill, it mandates security best practices including making paths relative and performing validation in independent directories to prevent directory traversal or dependency leaks.
Audit Metadata