tooluniverse-polygenic-risk-score
Pass
Audited by Gen Agent Trust Hub on Mar 15, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill's instructions and documentation were analyzed for all 10 threat categories, including prompt injection, obfuscation, and malicious persistence. No such patterns were detected.
- [DATA_EXFILTRATION]: The skill describes processes for handling sensitive genetic information (e.g., 23andMe raw data, VCF files). The analysis confirmed that these data are used for intended local calculations with no instructions or tools present for exfiltrating this data to unauthorized external servers.
- [EXTERNAL_DOWNLOADS]: All external URLs and resources referenced in the skill point to well-known and trusted academic or research institutions, such as the EMBL-EBI GWAS Catalog, Open Targets Genetics (Wellcome Sanger Institute/EMBL-EBI), and the Broad Institute. These are categorized as safe sources under the analysis guidelines.
- [PROMPT_INJECTION]: The skill provides standard instructional triggers and does not contain any attempts to bypass agent safety filters, override system instructions, or extract system prompts.
Audit Metadata