ab-test-setup
Pass
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: LOWSAFE
Full Analysis
- Prompt Injection (SAFE): No attempts to override agent safety filters or system instructions were detected. The 'Non-Negotiable' principles are standard instructional constraints for the A/B testing methodology.- Data Exposure & Exfiltration (SAFE): No access to sensitive file paths, hardcoded credentials, or unauthorized network operations were identified.- Unverifiable Dependencies & RCE (SAFE): The skill contains no package installations or remote script execution patterns.- Indirect Prompt Injection (LOW): The skill processes external user content (hypotheses, metrics, and traffic data). While it lacks explicit boundary markers to delimit user input, the skill has no 'write', 'execute', or 'network' capabilities, limiting the potential impact to the reasoning within the current session.- Dynamic Execution (SAFE): There is no evidence of runtime code generation, compilation, or unsafe deserialization.
Audit Metadata