The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external genomic data files, creating a surface for indirect prompt injection.
Ingestion points: Data is ingested from BED, FASTA, and fragment TSV files through modules such as RegionSet, RefgetStore, and TreeTokenizer.
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are specified for the processed data.
Capability inventory: Capabilities include file system write operations (to_bed, to_json, to_bigwig) and command-line tool execution (gtars-cli).
Sanitization: The documentation does not describe any sanitization or validation procedures for the content of the ingested genomic files.
[EXTERNAL_DOWNLOADS]: The skill involves the installation of external software and the retrieval of remote data.
It provides instructions to install the gtars Python package and the gtars-cli Rust-based tool.
The bbcache module is designed to fetch BED files from the bedbase.org bioinformatics database.
[DATA_EXFILTRATION]: The skill performs network operations to fetch data from an external genomic resource.
The bbcache fetch command initiates network requests to bedbase.org to download genomic files, which is a domain-specific operation for this toolset.

gtars