The Agent Skills Directory

[SAFE]: The skill is designed for scientific benchmarking of interatomic potentials. The logic is transparent and follows standard practices for materials science data analysis.\n- [EXTERNAL_DOWNLOADS]: The fetch_r2scan.py script retrieves structural and energy data from the Materials Project API. This is a well-known, legitimate data source for the materials science community.\n- [COMMAND_EXECUTION]: The instructions utilize conda run to manage Python environments for different model backends (MACE, MatGL, etc.). This is a standard procedure for ensuring reproducible environments in machine learning workflows.\n- [SAFE]: The skill uses pymatgen and mp-api, which are authoritative and trusted libraries in the field of computational materials science.\n- [SAFE]: Absolute local file paths are present in documentation (e.g., /home/bdeng/projects/...). While this exposes information about the author's local environment, it does not represent a functional security risk.

ml-mlip-benchmark