build-game-monster-system
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and interpret external monster definitions and asset sources, which introduces an attack surface for indirect prompt injection where instructions could be hidden in metadata or authored data fields.
- Ingestion points: Ingests
MonsterDefinitionauthored facts, tags, and asset source paths/URLs (SKILL.md, references/monster-contract.md). - Boundary markers: The instructions do not define explicit boundary markers or isolation protocols for data loaded from external asset contracts.
- Capability inventory: The skill focuses on character rigging, animation states, and LOD management; analysis shows no capabilities for subprocess execution, network exfiltration, or sensitive file system access.
- Sanitization: The skill implements a "Verify conformance" section and deterministic fixtures to validate that asset behavior matches the runtime contract, providing a level of functional sanitization.
Audit Metadata