deepchem
Pass
Audited by Gen Agent Trust Hub on Apr 9, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is built around the well-established 'deepchem' library, a standard tool for molecular informatics and machine learning in chemistry. Analysis of the scripts and documentation reveals legitimate scientific functionality with no malicious intent.\n- [EXTERNAL_DOWNLOADS]: The skill references downloads for benchmark datasets (MoleculeNet) and pre-trained transformer models from Hugging Face (such as ChemBERTa). These downloads are performed through official library APIs from trusted organizations and are necessary for the skill's primary function.\n- [COMMAND_EXECUTION]: The provided scripts in the 'scripts/' directory are well-structured command-line utilities for common ML tasks. They use standard argument parsing and do not contain any unsafe subprocess calls, shell injections, or hidden commands.\n- [DATA_EXFILTRATION]: The skill processes scientific data formats (SMILES strings, SDF, FASTA). It does not access sensitive system files, environment variables, or credentials. Data handling is limited to the files explicitly provided for analysis or benchmark datasets.
Audit Metadata