jaredrhod/ai-visualizer ? reverse-engineered prompt

Reverse engineered prompt

Build me this AI visualizer so I can run it locally and use it on stream.

I want a simple folder based web app that opens in a browser, starts with a tiny Python server, and shows a gallery with four full screen visual faces. One should be a living circuit board, one a particle starburst orb, one Matrix style rain with a face in it, and one a constellation brain. It should respond live to an AI voice line through the little signal files the README mentions, and it should also have a demo mode so I can see it working without connecting anything.

Please make it easy to configure my agent name and default face, and let me swap in my own portrait for the rain face. If anything is missing, use the README and current docs online to fill in the gaps. I want it to work nicely for OBS too, so each face can be opened as its own page.