tooluniverse-precision-medicine-stratification
Pass
Audited by Gen Agent Trust Hub on Apr 16, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill follows a highly structured workflow for clinical decision support, referencing well-known and trusted technology services and technology companies in the medical field, including NCBI, Ensembl, and the FDA. All external references are appropriate for the stated purpose.
- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface typical of agents processing user-supplied data. User-provided disease names and genomic profiles are passed as parameters to external tools. However, the structured nature of these tool calls and the requirement for specific medical identifiers (EFO/MONDO IDs) significantly mitigate the risk of cross-domain instruction leakage.
- Ingestion points: User-provided disease descriptors, genomic variants, and clinical biomarkers (SKILL.md).
- Boundary markers: Not explicitly defined; instructions rely on structured parsing of inputs.
- Capability inventory: Extensive list of external medical and genomic tools for data retrieval and analysis.
- Sanitization: The instructions mandate normalization and resolution of gene and disease IDs (Phase 1), which serves as a validation step before further processing.
Audit Metadata