gget

Pass

Audited by Gen Agent Trust Hub on Feb 15, 2026

Risk Level: LOWEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS] (INFO): The scripts rely on the gget library to perform network requests to external biological databases (e.g., NCBI for BLAST, Ensembl for gene info, and ARCHS4 for expression data). These are standard operations for bioinformatics tools.
  • [PROMPT_INJECTION] (LOW): Category 8: Indirect Prompt Injection surface. The skill ingests untrusted data from FASTA files (read_fasta in batch_sequence_analysis.py) and gene lists (read_gene_list in enrichment_pipeline.py).
  • Ingestion points: FASTA headers and sequence IDs, CSV columns containing gene names.
  • Boundary markers: None. Data is parsed directly into Python dictionaries and dataframes.
  • Capability inventory: File system writes (to_csv, f.write), network operations (via gget API calls).
  • Sanitization: Standard biological format parsing is performed, but no specific sanitization against prompt-injection-style strings (e.g., instructions hidden in FASTA headers) is present. If the agent later processes the resulting CSV or FASTA outputs, it could be influenced by malicious strings within those files.
Audit Metadata
Risk Level
LOW
Analyzed
Feb 15, 2026, 10:16 PM