product-definition-builder
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill and its associated scripts follow defensive design patterns, focusing on structure validation and owner-approved documentation lifecycle management.
- [COMMAND_EXECUTION]: Local system commands are executed exclusively for internal verification and unit testing. The Git integration in
scripts/git_evidence.pyis specifically hardened to prevent command injection, binary spoofing, and environment-based repository identity overrides. - [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user requirements to generate PRDs, presenting a standard attack surface for indirect injection. This is mitigated by mandatory structural validation using provided Python scripts, required machine-anchored approval markers, and 'Research Disclosure Checks' designed to prevent exfiltration of sensitive context during automated research tasks.
- [EXTERNAL_DOWNLOADS]: Reference material is cited from well-known technology organizations and cloud providers (e.g., Cloudflare, OpenAI, Vercel, Stripe), but the skill does not perform runtime downloads or execution of external payloads.
Audit Metadata