ai-feature-prd
Pass
Audited by Gen Agent Trust Hub on Oct 25, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill consists entirely of markdown templates and documentation. It contains no executable scripts, automated tools, or binary files.
- [PROMPT_INJECTION]: Heuristic detections flagged phrases like 'ignore previous instructions' in the reference and taxonomy files. These instances are strictly examples of adversarial inputs meant to guide PMs in designing red-team tests for their own features. They do not constitute an attempt to inject instructions into the current agent.
- [DATA_EXFILTRATION]: The skill includes best-practice advice on identifying and mitigating PII leakage and cross-tenant data exposure in AI products. It does not perform any data access or network operations itself.
- [EXTERNAL_DOWNLOADS]: The skill references several well-known third-party tools for AI evaluation and observability (Promptfoo, Langfuse, RAGAS, etc.). These are industry-standard tools and their mention is appropriate for the skill's domain.
Audit Metadata