bio-atac-seq-differential-accessibility
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill represents a surface for indirect prompt injection as it processes external genomic data and sample metadata.
- Ingestion points: Genomic data files (BAM, narrowPeak) and the sample_sheet CSV file processed in examples/diff_accessibility.R.
- Boundary markers: None present; the skill assumes the integrity of the input files.
- Capability inventory: The R script performs file reads and generates output files (CSV, PDF, BED) using standard library functions but lacks network or arbitrary shell execution capabilities.
- Sanitization: No explicit sanitization or validation of input file contents or metadata is performed beyond standard package loading.
- [SAFE]: Analysis of the instructions and R code confirms that all dependencies are standard Bioconductor or CRAN packages. No evidence of credential exfiltration, remote code execution, or persistence mechanisms was found. The use of external tools like MACS2, Subread, and Bedtools via Conda is standard practice for this domain.
Audit Metadata