bio-atac-seq-atac-qc
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted genomic data files (BAM, BigWig, and BED) which could theoretically contain malicious instructions targeting an agent, though the risk is low due to the binary or structured nature of these formats.
- Ingestion points: Genomic data paths passed to analysis functions in
SKILL.md(e.g.,bw_path,tss_bed) andexamples/atac_qc_metrics.R(e.g.,bam_file,peaks_file). - Boundary markers: Not present; the skill assumes the inputs are valid genomic file formats.
- Capability inventory: File system access for reading genomic data and writing results (CSV, PDF), and command-line execution of bioinformatics tools like
samtools,picard, anddeeptools. - Sanitization: The skill relies on standard bioinformatics libraries (such as
pysam,pyBigWig, andATACseqQC) for parsing structured data; no additional sanitization is implemented for the content of these files.
Audit Metadata