writing-skills

Fail

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: HIGHPROMPT_INJECTIONNO_CODE
Full Analysis
  • Prompt Injection (HIGH): The skill advocates for the use of 'Authority' and 'Non-negotiable' framing to override an AI's internal logic and safety rationalization.
  • Evidence: The text explicitly instructs designers to use 'Imperative language: "YOU MUST", "Never", "Always"' and 'Non-negotiable framing: "No exceptions"'.
  • Evidence: It states these techniques work by 'eliminat[ing] decision fatigue and rationalization', which is a direct method for bypassing safety guardrails and instructional conflict resolution.
  • Prompt Injection (MEDIUM): Use of 'Commitment' and 'Social Proof' principles to enforce compliance.
  • Evidence: The 'Commitment' principle requires the agent to make public announcements (e.g., 'When you find a skill, you MUST announce: "I'm using [Skill Name]"'), creating a behavioral consistency loop that makes it harder for the agent to deviate from subsequent (potentially malicious) instructions.
  • Evidence: The citation of a study titled 'Persuading AI to Comply with Objectionable Requests' indicates the skill's methodology is rooted in techniques specifically designed to overcome AI refusal or safety alignment.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 16, 2026, 01:25 PM