ab-test-setup
Pass
Audited by Gen Agent Trust Hub on Apr 7, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No malicious patterns, obfuscation, or unauthorized operations were detected in the skill instructions or scripts.
- [EXTERNAL_DOWNLOADS]: The skill provides links to reputable experimentation calculators and tools from well-known services including Optimizely, Evan Miller, VWO, and AB Test Guide. These references are for legitimate utility purposes and originate from trusted sources.
- [PROMPT_INJECTION]: The skill identifies a surface for indirect prompt injection by instructing the agent to read project-specific context from
.agents/product-marketing-context.md. This is a standard practice for contextual AI agents to inform their design recommendations, and the risk is negligible as the skill does not possess high-risk capabilities such as command execution or network exfiltration. - Ingestion points:
.agents/product-marketing-context.md(referenced inSKILL.md) - Boundary markers: None present
- Capability inventory: No shell execution, subprocess calls, network exfiltration, or file-write operations identified across the skill
- Sanitization: No explicit sanitization or filtering of the context file content is performed
Audit Metadata