meeting-prep-debrief
Pass
Audited by Gen Agent Trust Hub on Oct 25, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill utilizes a sophisticated 'Shared Core' operating principles architecture that prioritizes rigor, fact-verification, and commercial safety across all workflows.
- [SAFE]: Implements a strict 4-tier source hierarchy for factual claims, explicitly forbidding reliance on low-quality or unverifiable Tier 4 sources such as random blogs or AI-generated comparison sites.
- [SAFE]: Includes a 'Challenge-Assumptions' rule that mandates automated and manual verification of competitor relationships, URL existence, and regulatory contexts before generating deliverables.
- [SAFE]: Enforces a 'No-Harmful-Output' rule that specifically prevents commercial 'own-goals', such as generating competitive materials against products already owned by the user's company due to M&A activity.
- [INDIRECT_PROMPT_INJECTION]: The skill possesses a theoretical attack surface because it ingests untrusted data (attendee names, meeting descriptions, and LinkedIn content). However, this risk is effectively mitigated by the mandatory 4-tier verification protocol and the Human-In-The-Loop (HILT) Question Budget, which requires manual confirmation for ambiguous or unverified data.
Audit Metadata