tooluniverse-gwas-finemapping
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from external scientific sources such as Open Targets Genetics (GraphQL) and the GWAS Catalog (REST API). While this creates an ingestion surface for untrusted data (e.g., study metadata or variant annotations), these are well-known and reputable scientific databases. The skill uses this data to prioritize variants and link them to genes, which is consistent with its stated purpose. Standard LLM guardrails are sufficient for this scientific use case.
- [EXTERNAL_DOWNLOADS]: The documentation references external scientific services including Open Targets and the GWAS Catalog. These are established bioinformatics resources used for legitimate research purposes. No suspicious or unverified third-party domains were detected.
- [COMMAND_EXECUTION]: The skill provides Python examples that utilize a local module named
python_implementation. These examples demonstrate standard data processing workflows for genetic analysis and do not include arbitrary command execution or shell injection patterns.
Audit Metadata