unity-cleaner
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill scans and processes project assets (materials, textures, etc.) which are untrusted external inputs.
- Ingestion points: Local Unity project files accessed via the 'searchPath' parameter in scanning tools.
- Boundary markers: No explicit delimiters or instructions are used to separate asset content from agent instructions.
- Capability inventory: The skill has the ability to delete files (cleaner_delete_assets) and modify components (cleaner_fix_missing_scripts).
- Sanitization: The instructions do not specify any validation or sanitization of asset metadata or content.
- [SAFE]: Destructive operations such as asset deletion require a 'two-step confirmToken handshake', ensuring that a preview step is performed before any actual deletion occurs. The skill also explicitly documents its integration with platform-level operating modes to enforce authorization for dangerous actions.
Audit Metadata