tune-enemy-ai
Pass
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [SAFE]: The skill consists of high-level architectural guidelines for game development (state machines, perception-intent-motion separation, and testing fixtures). No executable scripts or shell commands are present in the analyzed files.
- [INDIRECT_PROMPT_INJECTION]: The skill is intended to process and refine logic for enemy AI behaviors. This creates a potential surface where untrusted external data (such as enemy behavior scripts or design documents provided by a user) could contain embedded instructions.
- Ingestion points: External AI behavior definitions and logic designs processed by the agent in SKILL.md context.
- Boundary markers: None provided in the instructions.
- Capability inventory: No scripts or tool usage defined in the skill files, though the agent's default capabilities would apply.
- Sanitization: No specific sanitization or filtering of external input is described.
Audit Metadata