tooluniverse-cancer-variant-interpretation
Pass
Audited by Gen Agent Trust Hub on Mar 15, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill integrates with multiple well-known and trusted biomedical services, including PubMed (NIH), the FDA, UniProt, and OpenTargets. These are authoritative sources in the medical domain and are accessed through predefined tools.
- [SAFE]: No prompt injection or behavior override patterns were detected. The instructions focus on tool parameter verification and structured report generation.
- [SAFE]: The skill does not perform any sensitive file access, credential exfiltration, or unauthorized network operations. All network activity is limited to the documented tool-calling workflow for data retrieval.
- [SAFE]: No code or script obfuscation techniques (such as Base64, hex encoding, or hidden characters) were found in the skill definitions.
- [SAFE]: The skill does not execute external scripts or install unverifiable dependencies. The provided Python snippets are illustrative instructions for the agent on how to use internal toolsets.
- [SAFE]: Indirect prompt injection risk is evaluated as minimal. While the skill ingests data from external APIs (PubMed, CIViC), the lack of dangerous capabilities like shell execution or arbitrary file writing prevents common exploitation paths.
Audit Metadata