bio-pathway-gsea
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests external datasets (gene expression results) which provides a theoretical surface for indirect prompt injection if data files were to contain malicious instructions.
- Ingestion points: Data is loaded from external files using
read.csv('de_results.csv')inSKILL.mdand referenced R scripts. - Boundary markers: The instructions do not define strict boundary markers or include specific 'ignore instructions' warnings for the data being processed.
- Capability inventory: The skill has the capability to perform network operations (via the live KEGG REST API) and write results to the file system (using
write.csvto temporary directories). - Sanitization: No specific filtering or sanitization of the input CSV data is performed to detect or block natural language instructions.
- [EXTERNAL_DOWNLOADS]: The skill connects to well-known external biological databases and software registries.
- The
gseKEGGfunctionality performs live queries to the official KEGG REST API to retrieve the most recent pathway annotations. - The usage guide provides instructions for downloading and installing necessary R packages from the official Bioconductor and CRAN registries.
Audit Metadata