tooluniverse-gwas-drug-discovery

Pass

Audited by Gen Agent Trust Hub on Mar 29, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill facilitates drug discovery by querying well-known and reputable scientific and regulatory databases, including the GWAS Catalog, Open Targets, ChEMBL, and the FDA. No malicious code, data exfiltration, or persistence mechanisms were detected.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes data from multiple external scientific sources which constitutes an indirect prompt injection surface. [Ingestion points]: Tool outputs from genomic and clinical databases in 'python_implementation.py'. [Boundary markers]: None; tool outputs are processed directly into research recommendations. [Capability inventory]: Restricted to specialized domain-specific tool calls; no broad shell or file system access was identified. [Sanitization]: None. This surface is assessed as safe given the reputable nature of the target sources and the intended scientific use case.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 29, 2026, 11:47 PM