PrefectHQ/prefect — reverse-engineered prompt

Build me a Python workflow tool that turns normal scripts into reliable data pipelines. I want users to mark functions as tasks and group them into flows, then run them locally or on a schedule without needing a lot of setup.

It should handle retries when something fails, simple caching so repeated work can be skipped, task dependencies, branching logic, and clear logs for every run. Add a local server with a clean web dashboard where I can see recent workflow runs, their status, logs, timing, and errors. I also want to trigger a run manually from the dashboard or from the command line.

Please include a small example workflow that checks GitHub repo stars, plus instructions for installing, starting the server, opening the UI, and scheduling the example to run every minute. Keep the first version focused and usable for a solo data person running pipelines on their laptop. Look up current docs online if you need to.