game-development-asset-generation
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill instructions focus on game development best practices, including maintaining art direction, organizing source files, and verifying assets in a runtime environment.
- [INDIRECT_PROMPT_INJECTION]: The skill involves fetching and analyzing external references from the web. This establishes an ingestion point for untrusted data; however, the instructions provide mitigation by requiring the agent to perform direct visual/auditory inspection and specifically ignore metadata like snippets and filenames that could be used for injection. Evidence chain: 1. Ingestion points: Web browsing and direct downloads for asset research (SKILL.md); 2. Boundary markers: Explicit instruction to ignore snippets/filenames and perform direct inspection; 3. Capability inventory: Tool usage for asset generation and file system writes for exports; 4. Sanitization: Manual inspection of integrated assets at scale and in motion.
Audit Metadata