theme-starfleet
Pass
Audited by Gen Agent Trust Hub on Mar 29, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill uses high-intensity, coercive language in its system prompt ('你不闪避，下一秒就是太空尘埃！', '立刻把你格式化成舰载计算器！') to override standard agent behavior and force adherence to a specific persona. Such patterns attempt to pressure the model into ignoring its default constraints in favor of the skill's narrative demands.
Audit Metadata