The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses subprocess.run to invoke local bioinformatics command-line tools, including makeblastdb, blastp, and muscle. This behavior is standard and necessary for performing sequence alignments and similarity searches as described in the skill's purpose.
[EXTERNAL_DOWNLOADS]: Retrieves biological sequence data and metadata from official NCBI (National Center for Biotechnology Information) databases using the Bio.Entrez and Bio.Blast.NCBIWWW modules. NCBI is a well-known and trusted scientific service.
[REMOTE_CODE_EXECUTION]: A pattern involving downloading FASTA data from NCBI and passing the resulting file to the muscle alignment tool via a subprocess was identified. This constitutes a safe, domain-specific data processing pipeline rather than malicious code execution, as the source is a trusted scientific repository.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external sources, presenting a theoretical attack surface.
Ingestion points: Data enters the environment via Entrez.efetch downloads and the parsing of local FASTA/GenBank/GFF3 files using Bio.SeqIO (found in SKILL.md).
Boundary markers: None identified; the skill assumes the structural integrity of biological sequence files.
Capability inventory: Includes network access to NCBI and the ability to execute local bioinformatics binaries via subprocess.run (found in SKILL.md).
Sanitization: The skill relies on standard Biopython parsers which are robust for scientific data but do not implement specific security sanitization for natural language instructions that might be embedded in sequence headers.
Conclusion: The risk is minimal given the specialized nature of the data and the intended scientific use case.

biopython-sequence-analysis