scenario-blender-previs-storyboard
Pass
Audited by Gen Agent Trust Hub on Sep 26, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: No malicious patterns, obfuscation, or unauthorized data access were detected. The skill is highly specialized for Blender production workflows and adheres to safe coding practices within that environment.
- [EXTERNAL_DOWNLOADS]: The skill instructions mention installing sibling skills from the vendor 'scenario-labs' using the platform's 'npx' command. This is a standard and legitimate way to manage dependencies within the specific agent platform.
- [COMMAND_EXECUTION]: The script uses Blender's internal Python API (bpy.ops) to perform rendering and scene manipulation. These operations are restricted to the local Blender instance and are necessary for the skill's purpose.
- [INDIRECT_PROMPT_INJECTION]: The skill has a surface for indirect prompt injection as it processes screenplay 'beats' to generate scene data. 1. Ingestion points: The BEATS list variable processed in scripts/bx_previs.py. 2. Boundary markers: Absent. 3. Capability inventory: Internal Blender rendering, camera placement, and mesh manipulation. 4. Sanitization: The skill relies on the Blender data model which treats inputs as metadata or 3D properties rather than executable commands.
Audit Metadata