Roy3838/Observer — reverse-engineered prompt

Build me a privacy first app called Observer AI where I can create little AI agents that watch what is happening on my computer and react for me. I want a simple dashboard to create and manage agents, give each one a prompt, pick a loop interval, choose inputs like screenshots, screen OCR, camera, clipboard, mic, and screen audio, then let it use local AI models to decide what to do.

The agents should be able to log text and images to memory, record clips, label important moments, and send alerts through email, Discord, Telegram, WhatsApp, SMS, phone call, browser notifications, and simple in app popups. I also want a code area for custom logic after each model response, with helper functions for memory, starting and stopping agents, time, notifications, and recordings.

Please make it feel polished and easy for non technical users, with a web app and a desktop app experience if that fits the repo. Keep user data local and secure by default. Look up current docs online if you need to.