LostRuins/koboldcpp ? reverse-engineered prompt

Build me a simple app that lets me run local AI models on my own computer without any setup pain. I want to drop in a compatible model file, launch one executable, and get a web UI where I can chat with it, do story or roleplay style conversations, and tweak a few settings like speed and GPU usage. It should work on Windows, Linux, and Mac if possible, and also be able to run on CPU when I don’t have a good GPU.

Please include a clean built in browser interface, a local server I can open at localhost, and support for common AI endpoints so other tools can connect to it too. If there are useful extras like image, speech, or vision features in the repo, wire those in as well. Keep it beginner friendly, with sensible defaults and a way to load models easily. If you need to look up current docs online while wiring it up, go ahead.