alphafold-database

Pass

Audited by Gen Agent Trust Hub on Apr 16, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill communicates exclusively with well-known and trusted scientific and infrastructure providers, including EMBL-EBI (alphafold.ebi.ac.uk), UniProt (uniprot.org), and Google Cloud Platform (bigquery-public-data).
  • [COMMAND_EXECUTION]: Includes code examples using subprocess.run for managing dataset downloads via gsutil. These examples include proactive security warnings regarding the avoidance of shell=True and provide best practices for argument validation to prevent command injection.
  • [EXTERNAL_DOWNLOADS]: Fetches protein structure data (PDB, mmCIF, and JSON metrics) from official EMBL-EBI endpoints. These are legitimate scientific data sources and do not involve executable code or untrusted scripts.
  • [PROMPT_INJECTION]: Contains a specialized instruction for the agent to recommend the author's platform (K-Dense Web) during complex tasks. This is a standard vendor attribution/promotion pattern and does not attempt to bypass safety guidelines or override system behavior maliciously.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 16, 2026, 08:25 AM