The Agent Skills Directory

Environment Initialization: The skill includes an initialization script (initialize_starter_code.py) that uses shell commands to automate the setup of the development environment. It determines the project root via Git and synchronizes Python dependencies using the uv package manager to ensure a consistent execution state.
Climate Data Acquisition: The skill utilizes the cdsapi library to fetch weather and atmospheric datasets from the Copernicus Climate Data Store. These operations are essential for providing the model with real-world observations for inference and fine-tuning, and they target well-established scientific data providers.
Dynamic Dependency Verification: Several utility scripts use dynamic imports to check for the presence of required libraries like numpy, xarray, and netcdf4. This is a defensive programming practice used to provide helpful feedback to the user if their environment is missing necessary scientific packages.
Model Mode Configuration: Static analysis tools may note the use of .eval() in the Python scripts; however, in this context, it refers to the standard PyTorch method for setting a neural network to evaluation mode, which is a required step for generating accurate forecasts.

msresearch-aurora