dagster-io/dagster ? reverse-engineered prompt

Build me a Python data pipeline orchestration tool like Dagster, where I can define data assets as normal Python functions and the system figures out the order to run them. I want to be able to declare dependencies between assets, run them on a schedule or on demand, and see which assets feed into which others.

Please also include a simple web UI that shows the asset graph, run history, and basic observability like success, failure, and logs so I can understand what happened when a pipeline breaks. It should feel friendly for local development, testing, and production use, not just a toy script.

Keep the design declarative and easy to test, and make it work with common Python data workflows. If you need to look up current docs online to match modern best practices, go ahead.