Leopold Wicht shared you an appI want to run ComfyUI locally using modal.com, by converting ComfyUI into a serverless service, wrapping the current API and having it be a modal function that ca be run by calling `modal serve comfy.py`. I want it to build with models and plugins already pre-installed, as defined by a list of my creation. The models could come from huggingface or civitai, an the plugins from github. Make it fast for a cold boot, and have it only run when I do inference, having the creation time be "shut down". The instance should be intelligent enough to stay awake until the inference is done, after which it can shut down. User the modal.com API docs to do this.
