nazirlouis/ada — reverse-engineered prompt

Build me a desktop AI assistant called A.D.A. in Python that feels like a real time voice companion. I want a clean GUI with a text box, buttons to switch between webcam, screen share, and no video, plus an animated avatar that reacts when it talks.

It should let me talk to it naturally, type messages if I want, and use live visual input from either my camera or my screen so it can answer questions about what it sees. I also want it to be able to do useful actions like search the web, run and debug Python code, read and edit files on my computer, and open apps or websites when asked.

Use Gemini for the assistant brain and ElevenLabs for speech, with API keys loaded from a .env file. Make it work on Windows, Mac, and Linux if possible. Please include the setup steps, required installs, and a simple way to launch it with camera, screen, or no video mode. If anything is unclear, look up the current docs online.