How do I download a specific checkpoint from CivitAI?

Use the download command: 'ujust comfyui download --model-url= --model-type=checkpoint'. The skill handles the placement into the correct subdirectory automatically.

How do I ensure my models are not deleted when the container restarts?

You must configure a persistent directory using the 'ujust comfyui config --models-dir=/your/path' command. By default, the skill runs in ephemeral mode where data is lost on container recreation.

Does this skill support NVIDIA GPUs?

Yes, it supports NVIDIA via CDI device passthrough. It also supports AMD via ROCm and Intel via oneAPI for hardware-accelerated image generation.

Can I install ComfyUI-Manager using this skill?

Absolutely. You can use the 'ujust comfyui node-install --node-url=https://github.com/ltdrdata/ComfyUI-Manager' command to install it or any other custom node repository.

ComfyUI Manager

Name: ComfyUI Manager
Author: atrawog

byatrawog

Data Science & ML

Deploys and manages ComfyUI instances for node-based Stable Diffusion image generation with GPU acceleration and model lifecycle support.

About

This skill provides a comprehensive interface for managing ComfyUI on Bazzite AI systems, enabling users to deploy, configure, and maintain a powerful node-based Stable Diffusion environment. It automates complex tasks such as container lifecycle management, GPU driver passthrough for NVIDIA, AMD, and Intel hardware, and persistent storage configuration. Users can streamline their creative workflows by managing custom nodes and downloading models directly from CivitAI using simple system commands, removing the technical friction of running local AI image synthesis.

Key Features

Flexible persistence configuration for models, outputs, and custom nodes
Direct model downloads from CivitAI with automatic directory categorization
0 GitHub stars
Automated container lifecycle management (start, stop, restart, delete)
Multi-vendor GPU support for NVIDIA (CDI), AMD (ROCm), and Intel (oneAPI)
Streamlined custom node installation and bulk update capabilities

Use Cases

Setting up a local, GPU-accelerated Stable Diffusion environment for high-quality image generation
Automating the download and organization of SDXL checkpoints, LoRAs, and VAEs from CivitAI
Managing complex node extensions like ComfyUI-Manager and ControlNet Aux through simple commands

About

Key Features

Flexible persistence configuration for models, outputs, and custom nodes
Direct model downloads from CivitAI with automatic directory categorization
0 GitHub stars
Automated container lifecycle management (start, stop, restart, delete)
Multi-vendor GPU support for NVIDIA (CDI), AMD (ROCm), and Intel (oneAPI)
Streamlined custom node installation and bulk update capabilities

Use Cases

Setting up a local, GPU-accelerated Stable Diffusion environment for high-quality image generation
Automating the download and organization of SDXL checkpoints, LoRAs, and VAEs from CivitAI
Managing complex node extensions like ComfyUI-Manager and ControlNet Aux through simple commands