# I hate ComfyUI's interface, so I vibe-coded a MOSS-Image frontend — retained record

> Retained reddit record from reddit:StableDiffusion.

Canonical: https://highsignal.app/data/records/f6090093f0c495a9

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:StableDiffusion)  published 2026-10-24 01:41 ingested 2026-10-24 10:59

[original source ↗](https://www.reddit.com/r/StableDiffusion/comments/1xdw30e/i_hate_comfyuis_interface_so_i_vibecoded_a/)

## retained text



```
Five days ago, I installed ComfyUI for the very first time because I wanted to try out the MOSS-Image model.

After seeing the workflow interface, my immediate thought was "I don't want to deal with this." So, instead of learning how to use ComfyUI, I just asked ChatGPT to make me a workflow. That then evolved into vibe-coding an entire browser based frontend that runs off of ComfyUI.

**And that's how MOSS Image Studio happened.**

**GitHub:** [https://github.com/roorem10/ComfyUI-MOSS-Studio](https://github.com/roorem10/ComfyUI-MOSS-Studio)

It's a browser-based frontend for MOSS-Image that still uses ComfyUI as the backend, but gives you a more traditional interface.

**Some of the features:**

* **Text-to-image generation** with MOSS-Image.
* **Experimental image-to-image** with image uploads and adjustable settings.
* **Model selection** for installed MOSS Base/SFT and INT8/BF16 variants.
* **Generation settings** including resolution, steps, guidance, sampler, scheduler, seeds, and batch count.
* **Advanced settings** for model-specific options.
* **Live progress tracking**, generation queue, and cancellation.
* **Image previews**, batch thumbnails, and downloads.
* **Save/load projects** and ComfyUI workflow exporting.
* **Persistent settings** so you don't have to reconfigure everything every time.

You can launch it through ComfyUI's Extensions menu or open it directly in your browser. It runs locally and doesn't require a separate frontend server. Installation instructions and requirements are in the GitHub README.

This project is **100% vibe-coded using ChatGPT**. The project works on my machine, but I can't guarantee it will work perfectly on everyone else's setup. Image-to-image is also experimental, so don't expect the same capabilities as a dedicated image-editing model. I figured I'd share this in case anyone else wants to use MOSS-Image without having to deal with the workflow interface.
```



document fetched 2026-10-25 00:30 · canonical https://reddit.com/r/StableDiffusion/comments/1xdw30e/i_hate_comfyuis_interface_so_i_vibecoded_a
