adversarial-verification

Pass

Audited by Gen Agent Trust Hub on Jul 25, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill provides instructions for a security audit workflow, which is a legitimate development and testing task.
  • [SAFE]: No malicious code execution, network exfiltration, or data exposure patterns were detected.
  • [SAFE]: The instructional persona ('Adversary') is scoped to the analysis of a target application and does not attempt to override the AI agent's own safety protocols or system instructions.
  • [SAFE]: No obfuscation, hidden content, or suspicious dependencies are present in the skill definition.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 25, 2026, 07:34 AM
Security Audit — agent-trust-hub — adversarial-verification