biology-biopython

Pass

Audited by Gen Agent Trust Hub on Sep 22, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill provides instructions for parsing external biological data formats (FASTA, GenBank, PDB, Newick). This creates an ingestion surface where malformed or specially crafted biological data files could potentially influence an agent's behavior if the data is treated as instructions.
  • Ingestion points: Ingestion occurs through SeqIO.parse, SeqIO.read, Phylo.read, and PDBParser.get_structure as described in SKILL.md.
  • Boundary markers: None are specified in the instructions to prevent the agent from interpreting sequence annotations or metadata as instructions.
  • Capability inventory: The skill documentation includes subprocess execution (Local BLAST), network access (NCBIWWW, Entrez), and file writing (SeqIO.write).
  • Sanitization: No explicit sanitization or validation of input data beyond standard Biopython parsing is mentioned.
  • [COMMAND_EXECUTION]: The skill mentions running Local BLAST via subprocess calls. This is a standard practice in bioinformatics for utilizing established command-line tools within a Python environment.
  • [EXTERNAL_DOWNLOADS]: The skill provides instructions for using NCBIWWW.qblast and Bio.Entrez to fetch data from NCBI databases. This represents legitimate network interaction with a well-known scientific service.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 22, 2026, 05:07 PM