tooluniverse-variant-analysis

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.80). The skill explicitly queries public third‑party databases (e.g., MyVariant.info, gnomAD, dbSNP, Ensembl/DGVa, ClinGen) as part of its workflow (see Phase 5 & Phase 7 in SKILL.md and references/annotation_guide.md and the ToolUniverse integration in python_implementation.py), and those external annotations are read and used to drive classification and reporting decisions—so untrusted third‑party content can materially influence the agent's actions.