The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external biological data files which could potentially contain malicious instructions if the data source is untrusted.
Ingestion points: Numerous file reading operations are present across SKILL.md and references/api_reference.md, including skbio.DNA.read(), skbio.io.read(), TreeNode.read(), DistanceMatrix.read(), and Table.read() for various formats (FASTA, FASTQ, Newick, BIOM, etc.).
Boundary markers: The instructions do not specify the use of delimiters or warnings for the agent to ignore embedded instructions within processed biological data.
Capability inventory: The skill provides file-writing capabilities across all primary data types, including seq.write(), tree.write(), table.write(), dm.write(), and pcoa_results.write() as seen in both SKILL.md and references/api_reference.md.
Sanitization: No explicit sanitization or validation of the content of the biological files is mentioned before the data is processed or potentially interpolated into further prompts.

scikit-bio