risk-management-specialist
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill metadata and core logic are aligned with professional regulatory standards and contain no prompt injection or behavioral bypass attempts.
- [SAFE]: The helper scripts provided are minimal placeholders that do not execute network requests, access sensitive files, or utilize dynamic execution.
- [INDIRECT_PROMPT_INJECTION]: The skill identifies ingestion points for external medical records and post-market data without implementing boundary markers or sanitization in the current skeletal files. The capability inventory includes placeholder scripts for risk analysis, representing a potential attack surface, although no active exploitation was observed.
Audit Metadata