torchdrug

Pass

Audited by Gen Agent Trust Hub on Jun 27, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is entirely instructional, providing documentation for the legitimate TorchDrug library. All provided code snippets are standard scientific Python examples for data loading, model definition, and training loops.
  • [EXTERNAL_DOWNLOADS]: The installation instructions use standard package managers (uv pip install torchdrug) to fetch the library from official package registries. The datasets mentioned (e.g., QM9, ZINC, Hetionet) are widely used academic benchmarks in the cheminformatics community and are accessed through the library's official dataset loaders.
  • [COMMAND_EXECUTION]: Shell commands included in the documentation are restricted to package installation and are provided as transparent examples for the user to follow, rather than automated scripts that would execute without oversight.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 27, 2026, 11:43 AM
Security Audit — agent-trust-hub — torchdrug