using-superpowers
Pass
Audited by Gen Agent Trust Hub on Feb 18, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- Prompt Injection (LOW): The instructions use extremely coercive language such as 'ABSOLUTELY MUST', 'not negotiable', and 'not optional' to force specific behaviors and override the agent's internal safety reasoning. It explicitly tells the agent that its reasoning is 'rationalization' that must be ignored. Evidence: 'You cannot rationalize your way out of this' (SKILL.md) and 'Stop — you're rationalizing' (references/red-flags-and-priority.md).
- Indirect Prompt Injection (LOW): The skill creates a dangerous surface by requiring tool invocation based on a '1% chance' threshold before the agent can gather context or ask questions. 1. Ingestion points: User messages. 2. Boundary markers: Absent. 3. Capability inventory: 'Skill' tool invocation. 4. Sanitization: Absent, as the skill directs the agent to act before clarifying or checking context.
Audit Metadata