paglagpt/LengtaMLX ? reverse-engineered prompt

Build me a macOS menu bar app for Apple Silicon that makes local LLMs easy to run and use from other apps. I want to point it at a folder of MLX model directories, see the available models in the menu bar, and start or stop them with one click. It should manage the model processes for me and keep track of which local model is available.

I also want a local API at http://127.0.0.1:2525/v1 that feels like a drop in replacement for OpenAI style tools, and also supports Anthropic style messages. If I send model=auto, it should choose the best local model automatically. If I use names like GPT, Claude, Gemini, Groq, Together, or OpenRouter models, it should route those to the right cloud provider when keys are configured.

Please include settings for model folder, Python environment check, API key, and provider keys, plus simple integration help for common coding tools. Make it feel polished and native on macOS. Look up current docs online if you need to.