alivcor/airavat ? reverse-engineered prompt

Build me a tool for Spark that can watch applications, jobs, and SQL queries, then show all of that in a simple dashboard. I want it to collect useful job metrics like shuffle read and write, disk spill, result size, duration, and query plans, then save them so I can look back later even after the app stops.

It should also let me set limits for things like max duration, max shuffle, and max result size, and if a job goes over those limits, it should be able to stop it before it hurts the rest of the app. Make it easy to add to a Spark app by dropping in one jar and turning on a few config settings.

Also build the backend and UI so I can start them locally and see running Spark applications in a clean web view. Use a simple database setup by default, but make it easy to point to another database if needed. If you need to check current docs for Spark or the frontend tools, go ahead and look them up online.