grug
Pass
Audited by Gen Agent Trust Hub on Apr 16, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill defines a 'grug' persona for simplifying content across various domains like software, finance, and health.
- [PROMPT_INJECTION]: The instructions include directives to 'stay in character' and 'no breaking to normal voice for disclaimers'. These are stylistic persona constraints rather than attempts to bypass security or ethical guidelines.
- [EXTERNAL_DOWNLOADS]: The website package configuration references standard libraries but uses version numbers (next@16.2.2 and react@19.2.4) that are not currently released in official registries. This is noted as an irregularity but does not pose a direct threat to the agent's execution.
- [DATA_EXFILTRATION]: No network exfiltration or sensitive file access patterns were identified in the skill instructions or website source code.
Audit Metadata