text-based-molecule-editing

Pass

Audited by Gen Agent Trust Hub on Mar 25, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill utilizes the open_biomed and rdkit libraries for standard cheminformatics tasks, such as property calculation (LogP, QED, SA) and SMILES processing.
  • [SAFE]: Model execution is performed via local checkpoints (./checkpoints/server/) using standard PyTorch-based inference pipelines, posing no risk of unauthorized remote code execution.
  • [SAFE]: All documentation and examples focus on scientific workflows. Troubleshooting advice, such as setting environment variables for mirrors or installing local packages in development mode, follows established developer practices.
  • [SAFE]: No obfuscation, data exfiltration patterns, or prompt injection vulnerabilities were identified within the skill's instructions or supporting scripts.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 25, 2026, 06:02 AM