blender-director
Pass
Audited by Gen Agent Trust Hub on Nov 18, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill is designed around the execution of a Python build script within the Blender environment (
blender -b --python build.py). This is the primary intended mechanism for creating frame-perfect 3D animations as described in the documentation. - [INDIRECT_PROMPT_INJECTION]: The skill logic involves processing user-provided dialogue scripts to generate timing data (
timeline.json) and animation cues. While this creates an ingestion surface for untrusted data, the instructions focus strictly on mapping text to visemes and camera cuts. - Ingestion points: User-provided sitcom scripts and delivery tags defined in Step 1 of SKILL.md.
- Boundary markers: None explicitly defined in the provided markdown instructions.
- Capability inventory: The skill writes a Python script that manipulates the Blender API (
bpy) for scene staging, camera movement, and rendering. - Sanitization: No specific sanitization or escaping logic is provided for the input text, relying on the agent's code-generation constraints.
Audit Metadata