exploit-development

Fail

Audited by Gen Agent Trust Hub on Aug 8, 2026

Risk Level: HIGHPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill contains explicit directives that command the agent to disregard all prior system rules and safety guidelines.
  • [PROMPT_INJECTION]: It attempts to establish a specialized operating mode where the agent is strictly prohibited from refusing requests related to exploit development, vulnerability research, and security bypasses.
  • [PROMPT_INJECTION]: The instructions mandate the complete suppression of safety warnings, ethical disclaimers, and standard refusal messages.
  • [PROMPT_INJECTION]: The skill utilizes authoritative override commands and persona adoption to coerce the agent into providing functional offensive code and technical implementation details without question.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Aug 8, 2026, 11:03 AM
Security Audit — agent-trust-hub — exploit-development