Home

MooshieUI Wiki

MooshieUI is a beginner-friendly interface for ComfyUI. It hides ComfyUI's node-graph complexity behind a clean, guided UI so you can generate without hand-editing graphs. It runs two ways:

• Desktop app via Tauri (Windows/Linux, macOS source build)
• Browser/server mode via a built-in web server (LAN/Docker friendly, mobile UI included)

Start here

• Installation - one-click setup, Docker, remote ComfyUI, macOS source build
• Generation Basics - the three modes and all the core controls
• Prompting Guide - wildcards, prompt presets, scheduling, regional prompts, artists/characters, autocomplete
• Prompt Assistant - the built-in local LLM that enhances or composes prompts

Features by area

• Upscaling and Face Fix - tiled diffusion upscaling, guidance nodes, face detailer
• ControlNet and Style Transfer - structural guidance and reference-image style transfer
• Inpainting and the Canvas Editor - multi-layer canvas, mask painting, staging strip
• Compare Grid - XYZ-style parameter sweeps with stitched output
• Models and the Model Hub - smart model detection, CivitAI/HuggingFace browser, model manager
• Gallery and Metadata - persistent gallery, PNG/JXL metadata, drag-drop remix
• Server, LAN and Multi-User - hosting MooshieUI for phones and other machines, roles and auth
• Settings and Accessibility - autocomplete config, vision simulator, theme, persistence
• FAQ

Notes on this wiki

This wiki documents user-facing behavior. For contributor/build docs see README.md, push-instructions.md, and CHANGELOG.md.