biopython-molecular-biology
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from untrusted external sources, including biological sequence files (FASTA, GenBank, FASTQ, PDB) and remote NCBI databases (PubMed, GenBank, Protein). This creates a surface for indirect prompt injection where malicious instructions could be embedded in sequence descriptions, publication abstracts, or other metadata fields.
- Ingestion points: Data is ingested through functions like
Bio.SeqIO.parse(),Bio.Entrez.read(),Bio.Blast.NCBIXML.read(), andBio.AlignIO.read()inSKILL.md. - Boundary markers: The skill does not implement explicit delimiters or instructions for the agent to ignore natural language found within the biological data.
- Capability inventory: The skill possesses capabilities to write files to the local system (
Bio.SeqIO.write(),Bio.Phylo.write(),pandas.DataFrame.to_csv()) and execute local command-line applications (Bio.Blast.Applications.NcbiblastpCommandline). - Sanitization: There is no evidence of sanitization or filtering applied to metadata or sequence descriptions before they are processed or output.
Audit Metadata