The Agent Skills Directory

[SAFE]: The skill functions as intended for cheminformatics tasks. The instructions and code examples are consistent with standard use of the datamol and RDKit libraries, and no indicators of malicious activity or security bypasses were found.
[EXTERNAL_DOWNLOADS]: The skill documentation instructs the agent to install the datamol package from the public registry using uv pip install datamol. This is a standard procedure for this type of tool.
[COMMAND_EXECUTION]: The skill requires permissions to execute Python and uv commands via the Bash tool to perform molecular computations and handle data files.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external molecular data from formats including SDF, CSV, and SMILES, creating an attack surface for indirect prompt injection.
Ingestion points: Data loading functions like dm.read_sdf(), dm.read_csv(), and dm.open_df() in references/io_module.md and SKILL.md.
Boundary markers: The skill documentation does not explicitly define prompt boundary markers for data content, relying on the structured nature of chemical file formats.
Capability inventory: The skill has the capability to write files (dm.to_sdf, dm.to_xlsx, dm.viz.to_image) and execute shell commands for Python scripts.
Sanitization: Standard chemical sanitization (dm.sanitize_mol) is applied to ensure structural validity, but not for prompt safety.

alterlab-datamol