The Agent Skills Directory

[SAFE]: No obfuscated code, hidden persistence mechanisms, or unauthorized data exfiltration attempts were found during the analysis of the skill files.
[EXTERNAL_DOWNLOADS]: The skill installs the well-known pysam Python package via uv pip. pysam is a standard bioinformatics library for working with high-throughput sequencing data.
[COMMAND_EXECUTION]: The skill provides access to samtools and bcftools commands through the pysam library wrapper. This functionality is intended for legitimate genomic file operations such as sorting, indexing, and format conversion.
[PROMPT_INJECTION]: The skill has a theoretical vulnerability to indirect prompt injection due to its processing of untrusted genomic data files (BAM, VCF, FASTA, etc.).
Ingestion points: Data is ingested from genomic alignment and variant files using pysam.AlignmentFile, pysam.VariantFile, and related classes (documented in references/alignment_files.md and references/variant_files.md).
Boundary markers: The skill does not define explicit delimiters or boundary markers to separate data from the processing context.
Capability inventory: The skill can execute bioinformatics subcommands via pysam.samtools and write new genomic files to the local system.
Sanitization: There is no evidence of sanitization for parsed metadata fields (e.g., read names, VCF header lines) before they are processed by the agent.

alterlab-pysam