Agent Data Ml Model

This is a specialized agent that handles the full ML workflow, from preprocessing data to training models and prepping them for deployment. It's scoped to work within data, models, and notebooks directories, with triggers on keywords like "train model" or "classification" and file patterns like .ipynb and .pkl files. The agent requires human approval before deploying models to production, which is smart. It comes with hooks that check for datasets and ML libraries before execution and reports on model artifacts after. The 30-minute execution limit and 100MB file size cap tell you it's meant for typical ML experiments, not massive distributed training jobs.