MooshieUI is a beginner-friendly interface for ComfyUI that runs in two modes:
- Desktop app via Tauri (Windows/Linux, macOS source build)
- Browser/server mode via the built-in web server (LAN/Docker friendly, mobile UI)
Built with Svelte 5 + Rust, it hides ComfyUI's node-graph complexity behind a clean, guided workflow so you can generate without hand-editing graphs.
Full guides live in the MooshieUI Wiki:
| Guide | Covers |
|---|---|
| Installation | Desktop, Docker, remote/cloud ComfyUI, macOS source build |
| Generation Basics | Modes, generation controls, dimensions |
| Prompting Guide | Autocomplete, presets, wildcards, interrogation |
| Prompt Assistant | LLM-assisted prompt building |
| Models & the Model Hub | Supported architectures, auto-detection, downloads |
| Upscaling & Face Fix | Tiled diffusion, guidance nodes, face fix |
| ControlNet & Style Transfer | ControlNet and reference/style transfer |
| Inpainting & the Canvas Editor | Mask painting and selective edits |
| Compare Grid | XYZ parameter sweeps |
| Gallery & Metadata | Persistent gallery, metadata import/remix |
| Server, LAN & Multi-User | Self-hosting, roles, auth, mobile |
| Settings & Accessibility | Persistence, i18n, accessibility |
| FAQ | Common questions |
- Three generation modes - text to image, image to image, and inpainting with a built-in canvas/mask editor; settings carry over between modes.
- Full generation controls - searchable checkpoint/VAE/LoRA pickers with auto-download, all ComfyUI samplers and schedulers, steps/CFG/seed/batch, and smart dimension presets.
- Smart model detection - 20+ architectures (SD 1.5, SDXL, Illustrious/NoobAI, Pony, SD3/3.5, the Flux family, Chroma, Z-Image, Wan, Qwen, AuraFlow, PixArt, HunyuanDiT, Stable Cascade, Kolors, Anima, Mugen, Nanosaur) identified by SHA256 hash, each with auto-applied sampler/scheduler/CFG presets.
- Tiled-diffusion upscaling - MultiDiffusion/SpotDiffusion with anti-hallucination guidance nodes, one-click upscale, and YOLOv8 face fix.
- Compare Grid (XYZ) - per-cell parameter sweeps stitched into a single labelled image.
- Real-time feedback - live latent previews, progress phases, and cancel, streamed over WebSocket.
- Gallery & metadata - persistent SQLite-backed gallery; drag a PNG back in to restore its settings (SwarmUI/A1111 metadata + stealth alpha).
- Self-hostable - headless web server with roles, per-user galleries, auth, and a dedicated mobile layout.
- 11 languages - 2,000+ translation keys with full locale parity, switchable without restart.
See the Wiki for the full feature reference.
- Download a release from Releases.
- Run the app. The setup wizard downloads uv, Python, ComfyUI, and PyTorch (GPU auto-detected) and installs MooshieUI's custom nodes - no Python or pip setup required.
- Start generating; ComfyUI launches automatically.
~5–10 GB disk, 5–15 minutes on first launch. macOS, Docker, and remote/cloud ComfyUI setups are covered in Installation.
cp .env.example .env # optional: credentials/ports
docker compose up -d --buildOpen http://localhost:3200 (or your configured MOOSHIEUI_PORT). Full server/LAN/multi-user setup: Server, LAN & Multi-User.
git clone https://github.com/Mooshieblob1/MooshieUI.git
cd MooshieUI
npm install
npm run tauri dev # hot-reload dev
npm run tauri build # production build- You adjust settings in the Svelte UI.
- On Generate, settings go to the Rust backend via the IPC bridge (
ipcInvoke()on desktop, HTTP/SSE in browser mode). - Rust builds a ComfyUI workflow JSON from templates - no node graph exposed.
- The workflow is submitted to ComfyUI's
/promptAPI. - WebSocket streams progress and previews back to the UI in real time.
MooshieUI also ships custom ComfyUI nodes (tiled diffusion, soft/smart guidance, an SDXL↔Flux2 VAE adapter, Nanosaur DiT support, and face fix) that are auto-installed into ComfyUI. Details live in Models & the Model Hub. The tiled diffusion node is also available as a standalone ComfyUI custom node: ComfyUI-MooshieTiledDiffusion.
| Layer | Technology |
|---|---|
| Frontend | Svelte 5, TypeScript 6, Tailwind CSS 4 |
| Runtime | Tauri desktop app + axum headless web server |
| State | Svelte 5 runes - class-based singleton stores |
| Persistence | Tauri Store (JSON) + SQLite (rusqlite) |
| ComfyUI transport | REST + WebSocket via Rust (reqwest, tokio-tungstenite) |
| Inference | ONNX Runtime (ort) for WD EVA02 image interrogation |
| Autocomplete | Danbooru + Anima tag databases (~140k tags) |
| i18n | 11 languages, 2,000+ keys, runtime switching |
| Build | Vite 6 + @sveltejs/vite-plugin-svelte |
Automated GlassWorm resistance checks run on every push and pull request to catch supply-chain attacks that hide payloads in invisible Unicode variation selectors or tamper with git timestamps. The CI workflow (.github/workflows/glassworm-scan.yml) blocks merges on failure. Contributors should enable the same checks locally:
bash scripts/setup-hooks.shPull requests are welcome. main is protected: open a PR from a chore/<topic> branch after local validation and GlassWorm pre-commit checks. See push-instructions.md for the full workflow (branch naming, build gates, IPC/gallery conventions, and CI).
See CHANGELOG.md for the full version history.
Licensed under the GNU Affero General Public License v3.0.
Built on ComfyUI, Tauri, Svelte, and Tailwind CSS. Uses CivitAI and HuggingFace for model data, Danbooru tags, OmniSR upscalers, YOLOv8 face detection, and Konva for the canvas editor. Tiled diffusion implements MultiDiffusion (Bar-Tal et al., ICML 2023) and SpotDiffusion (Ding et al., 2024).