biopython
Pass
Audited by Gen Agent Trust Hub on Oct 24, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill includes instructions for running external bioinformatics tools like Clustal Omega and MUSCLE using Biopython's command-line wrappers. These are standard operations within the bioinformatics domain and use documented library functions to interface with the local environment.\n- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion and parsing of various biological data formats, creating a potential surface for indirect prompt injection.\n
- Ingestion points: Biological data is parsed from external files using
SeqIO,AlignIO, andPDBParseracross various reference files (e.g.,references/sequence_io.md).\n - Boundary markers: The provided code examples do not include explicit prompt delimiters or instructions for the agent to ignore content within the data files.\n
- Capability inventory: The skill has the capability to execute shell commands via Biopython wrappers and perform network operations via NCBI Entrez and BLAST services.\n
- Sanitization: The skill relies on Biopython's internal parsing logic for validation; no additional sanitization of the extracted sequence data or metadata is mentioned before it enters the agent's context.
Audit Metadata