The Agent Skills Directory

[SAFE]: The skill provides a standard set of tools for medicinal chemistry, including drug-likeness rules (Lipinski, Veber), structural alerts (PAINS, NIBR, Lilly), and molecular complexity metrics. All referenced libraries ('medchem', 'datamol', 'rdkit') are well-known and legitimate tools in the chemoinformatics community.
[EXTERNAL_DOWNLOADS]: The skill instructs the installation of the 'medchem' package from PyPI ('uv pip install medchem'). This is a standard and safe dependency management practice.
[COMMAND_EXECUTION]: The provided script 'scripts/filter_molecules.py' is a legitimate utility for batch processing molecular data. It uses standard libraries for file I/O (pandas, pathlib) and molecular parsing (rdkit, datamol). No dangerous command execution or shell injection vectors were identified.
[DATA_EXFILTRATION]: No network operations or sensitive data access patterns were detected. The skill operates locally on provided molecular datasets.
[INDIRECT_PROMPT_INJECTION]: While the skill processes external data (SMILES strings and SDF files), this data is parsed as chemical structures by specialized libraries. There is no mechanism for the data content to influence agent instructions or bypass safety filters.

medchem