tooluniverse-precision-oncology
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFE
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references and fetches data from numerous well-known bioinformatics and medical services including PubMed, ClinicalTrials.gov, GDC (The Cancer Genome Atlas), UniProt, and ChEMBL. These are established repositories for clinical and genomic data.
- [COMMAND_EXECUTION]: Uses computational tools such as NvidiaNIM for AlphaFold2 structure prediction and docking analysis. These operations are performed within the scope of molecular research and do not involve arbitrary shell command execution.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests clinical molecular profiles (mutations, fusions) as untrusted data points. It mitigates risk by using structured tool calls for specific medical APIs (e.g., CIViC, ClinVar, OpenTargets) rather than passing raw input to a general-purpose execution environment.
- Ingestion points: Patient molecular profiles and clinical data provided by the user.
- Boundary markers: The skill uses structured markdown report templates to isolate generated clinical content.
- Capability inventory: Limited to medical database queries and structure-based computational tools via the
tu.toolsinterface. - Sanitization: Clinical queries are normalized through resolved gene identifiers and validated variant nomenclature before processing.
Audit Metadata