medchem

Pass

Audited by Gen Agent Trust Hub on Oct 24, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill implements standard medicinal chemistry functionality using established open-source libraries including medchem, datamol, and RDKit. All core filtering logic is executed locally.
  • [EXTERNAL_DOWNLOADS]: The documentation references an optional medchem install-lilly command. This command is a built-in feature of the medchem library used to download and compile native tools for specific demerit filters. This is documented as a standard extension of the library's functionality.
  • [INDIRECT_PROMPT_INJECTION]: The skill provides scripts/filter_molecules.py which ingests external molecular data (SDF, CSV, TXT). The script uses the RDKit and datamol libraries to parse chemical structures into internal objects for property calculation. There is no evidence that untrusted data from these files is interpolated into prompts or used to override agent instructions.
  • [COMMAND_EXECUTION]: The skill includes a utility script for batch processing. The script performs file I/O and mathematical computations based on user-provided arguments but does not facilitate arbitrary command injection.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 24, 2026, 09:29 PM