JackSpiece/OpenOmniBot — reverse-engineered prompt

Build me an Android app called OpenOmniBot, an on device AI assistant that does more than chat. I want it to understand what is on the screen, decide what to do, then actually use my phone like a person by tapping apps, doing gestures, changing settings, and handling simple Android tasks once permissions are granted. It should have a chat screen, a settings area for choosing AI providers and models, support memory for short term and long term context, and include useful tools like file access, a browser, a workspace, and a terminal.

I also want task automation features like scheduled jobs, alarms, calendar actions, audio control, and a way to run visual tasks from the chat page. Please include a skills system where I can paste a repo link to install a skill and then turn skills on or off. If possible, support both a lighter version and a fuller version with local model inference on device. Look up current docs online if you need to, but keep the app aligned with this repo and make it feel usable end to end.