The Agent Skills Directory

[COMMAND_EXECUTION] (SAFE): The skill includes documentation for invoking external bioinformatics tools (Clustal Omega and MUSCLE) through Biopython's command-line wrappers.
Evidence: ClustalOmegaCommandline and MuscleCommandline examples in references/alignment.md.
Context: These are standard library features for calling specialized external binaries, which is the primary purpose of this reference material.
[EXTERNAL_DOWNLOADS] (SAFE): The code snippets rely on the biopython package, a well-established and trusted library in the scientific community.
[PROMPT_INJECTION] (LOW): The skill demonstrates how to ingest data from external files (FASTA, GenBank, etc.), which creates a surface for indirect prompt injection if the processed sequence headers contain malicious instructions meant for an LLM.
Evidence Chain (Category 8):
1. Ingestion points: SeqIO.parse() and AlignIO.read() used across both files.
2. Boundary markers: None present in code snippets.
3. Capability inventory: File reading/writing and subprocess calls for sequence alignment.
4. Sanitization: No specific input sanitization is shown, though the library's parsers strictly validate biological sequence formats.

biopython