game-model-cleanup
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides instructions for the agent to execute shell commands that invoke the Blender executable with local Python tools provided by the vendor, theorcdev (e.g., tools/3d/audit_model.py, tools/3d/prepare_cleanup_model.py). These commands are central to the asset processing and validation workflow.
- [INDIRECT_PROMPT_INJECTION]: The skill exposes an attack surface for indirect prompt injection by processing external, potentially untrusted data. 1. Ingestion points: The workflow ingests data from external 3D models (FBX/GLB format), external JSON manifest files, and an external coordination board as described in SKILL.md and WORKFLOW.md. 2. Boundary markers: The skill does not define explicit delimiters or security markers to instruct the agent to ignore natural language instructions that might be embedded in the metadata of the 3D assets or manifest files. 3. Capability inventory: The skill possesses significant capabilities, including file system writing and shell command execution through the Blender environment. 4. Sanitization: While the workflow includes an 'audit_model.py' script, this tool is described as focusing on technical geometry and material validation rather than sanitizing inputs against malicious instructions.
Audit Metadata