blender-code-film
Pass
Audited by Gen Agent Trust Hub on Nov 14, 2026
Risk Level: SAFEDYNAMIC_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [DYNAMIC_EXECUTION]: The skill's primary function is to generate and execute a Python script (
scripts/film.py) via Blender's headless command-line interface (blender -b -P). This allows the agent to procedurally build 3D scenes, materials, and animations. - [COMMAND_EXECUTION]: The skill uses local shell access to execute
blenderfor rendering frames andffmpegfor video assembly and contact sheet generation via provided helper scripts (scripts/mux.shandscripts/sheet.sh). - [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied "briefs" to generate executable Python code, which constitutes an attack surface for indirect prompt injection.
- Ingestion points: User-provided text describing the film's look, shots, and behavior.
- Boundary markers: No specific delimiters are mandated in the prompt templates to separate user instructions from system guidelines.
- Capability inventory: The skill possesses the ability to
WriteandEditlocal files and executeBashcommands for Blender and media processing. - Sanitization: The skill advocates for a "critic loop" using a fresh subagent to audit images and drafts, which serves as a validation layer for the generated content.
Audit Metadata