perspective-reversal
Pass
Audited by Gen Agent Trust Hub on Oct 27, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill employs role-play instructions specifically designed to bypass the AI's standard 'cautious and objective' behavior. It directs the AI to 'pretend to be the adversary' and 'assume bad intentions' such as maximizing harm or exploiting the user. This pattern is a form of prompt injection that attempts to circumvent safety filters and neutrality constraints by reframing adversarial reasoning as a strategic intelligence exercise.
Audit Metadata