omnimind-ai/OpenOmniBot — reverse-engineered prompt

Build me an Android app called OpenOmniBot, an on device AI assistant that goes beyond chat and can actually help operate the phone.

I want users to chat with the assistant, configure their AI providers and models, grant permissions, then ask it to do phone tasks like open apps, tap through screens, use gestures, change settings, create or check calendar items, set reminders, control audio playback, browse the web, read and write workspace files, and use a terminal. It should have memory, both short term and long term, so it can remember useful context.

Please include a skill system where users can install skills from a repo link, enable or disable them, and run them from chat. Also include scheduled tasks and subagents for longer jobs. The app should support a standard version and a fuller version with local model inference using vision capable models when available.

Use native Android plus Flutter like this project does, and look up current docs online if needed.