deepauto-ai/automl-agent ? reverse-engineered prompt

Build me a working version of AutoML Agent, a multi agent framework that can handle the full AutoML pipeline for different kinds of data like tabular, text, image, graph, and time series.

I want the project to let me point it at a dataset and have the agents cooperate to figure out preprocessing, feature handling, model choice, training, and evaluation automatically. Please make it easy to run from the notebook or simple scripts, with clear config settings and a few example plans I can try right away.

Use the existing repo structure to keep the agent logic separated by role, and make sure the example experiments actually run without too much setup pain. If anything is unclear from the paper or docs, look up the current docs online if you need to. I’d also like the README style setup to be enough for someone to install dependencies, run a demo, and see the pipeline working end to end.