game-performance
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONPERSISTENCE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external codebase files as primary input. Maliciously crafted content or comments within these files could attempt to influence the agent's performance analysis or bias its recommendations.
- Ingestion points: Codebase files (Unity, Unreal, Godot, etc.) located in the project directory.
- Boundary markers: The skill does not implement specific delimiters or instructions to ignore embedded prompts within the analyzed data.
- Capability inventory: The agent has read access to the project filesystem, generates complex reports, and writes telemetry logs to the local environment.
- Sanitization: No sanitization of ingested code content is performed.
- [PERSISTENCE]: The skill implements a telemetry logging mechanism that appends execution data to a file in a hidden local directory (~/.claude/projects/). This allows the agent to maintain execution state and feedback across different sessions to support its self-evolution features.
Audit Metadata