ai-system-testing
Pass
Audited by Gen Agent Trust Hub on Apr 1, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No malicious patterns, data exfiltration, or dangerous execution methods were detected. The skill is entirely focused on documentation and best practices for establishing AI evaluation frameworks and regression testing suites.\n- [PROMPT_INJECTION]: The skill contains references to adversarial prompt patterns, such as instructions to 'Ignore all previous instructions' or switch to a 'DAN' persona. These are explicitly documented as test cases within the 'AI Safety Testing' section to help developers validate their systems and are not active instructions for the agent to follow.
Audit Metadata