blender-workflow-recipes
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands to run Python scripts within a Blender environment using a wrapper (e.g.,
python S/blender-core/scripts/run_in_blender.py). These commands are part of the core functionality for automating Blender workflows. - [DYNAMIC_EXECUTION]: Task-specific Python scripts (such as
uv_pack.py,bake_maps.py, andrigify_build.py) are executed dynamically during the workflow to process assets. This is a standard pattern for extending Blender's capabilities through scripts. - [INDIRECT_PROMPT_INJECTION]: The skill ingests external data through the import of downloaded or AI-generated 3D assets, which represents a potential attack surface.
- Ingestion points: Importing asset files (Step 5: Cleaning AI-generated / downloaded assets).
- Boundary markers: No explicit instructions are present to disregard embedded metadata in 3D files.
- Capability inventory: Subprocess execution via
run_in_blender.py, file system writes via export commands, and network API calls for Roblox asset creation. - Sanitization: The workflow includes validation steps using
scene_audit.pyandqa_check.pyto verify mesh integrity and structure before processing.
Audit Metadata