ab-test-setup
Pass
Audited by Gen Agent Trust Hub on Apr 29, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill instructions in
SKILL.mdand the reference materials in thereferences/directory focus on experimentation methodology and statistical rigor. No patterns of prompt injection, data exfiltration, or malicious commands were detected. - [SAFE]: External references to calculators (Evan Miller, Optimizely, VWO, etc.) are informational links to well-known industry services and do not involve automated data transfers or remote script executions.
- [SAFE]: The skill identifies and accesses local configuration files (e.g.,
.agents/product-marketing-context.md) to align its advice with the user's business context. This is a standard and safe operational pattern as the skill lacks capabilities for network communication or arbitrary file writing.
Audit Metadata