The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external biological data files (such as FASTA, FASTQ, and BIOM), which represents a potential attack surface for indirect prompt injection if those files contain malicious instructions.
Ingestion points: The skill utilizes several ingestion methods including skbio.DNA.read, skbio.io.read, Table.read, and TreeNode.read to process external files as described in both SKILL.md and references/api_reference.md.
Boundary markers: No specific boundary markers or instructions to disregard embedded commands in the processed data are provided in the skill documentation.
Capability inventory: The skill includes capabilities to write data back to the file system using various .write() methods (e.g., dna.write, table.write, dm.write), providing a path for potential unauthorized file modifications if the agent is compromised by injected data.
Sanitization: The documentation does not mention sanitization, validation, or escaping of the content read from external biological data formats.

scikit-bio