PotionUI

mcp
Security Audit
Fail
Health Pass
  • License — License: GPL-3.0
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Community trust — 65 GitHub stars
Code Fail
  • rm -rf — Recursive force deletion command in .github/workflows/backend-tests.yml
  • rm -rf — Recursive force deletion command in .github/workflows/docker-publish.yml
  • rm -rf — Recursive force deletion command in .github/workflows/docker-smoke.yml
Permissions Pass
  • Permissions — No dangerous permissions requested

No AI report is available for this listing yet.

SUMMARY

Self-hosted AI studio for images, video and music on your own GPU. Multi-user with logins and limits, no node graphs, drivable over MCP.

README.md
PotionUI

PotionUI

The self-hosted generation studio you can hand to other people.

Images, video and music on your own GPU · Free and open source · Linux & Windows

Latest release
Tests
License: GPL v3
Stars
Platform

Discord
Reddit
ko-fi

Install · Quick start · Website · Docs · Changelog · Contributing · Security

Run Krea-2, Anima, SDXL, Flux, Wan, LTX, Qwen-Image, MiniMax and YuE2 on one box — then give your team, your household, or your agents their own logins, presets, and limits. No node graphs. No per-model setup. Pick a model, type, watch it render live.

The generate workspace in three panes: the form folded to the side, a wide prompts pane holding one prompt for a potion bottle labelled ELIXIR, and the Krea-2 render beside it

Watch the 60-second tour (video)

https://github.com/user-attachments/assets/950415f7-da97-403e-811b-4c9c41d8106f

Why PotionUI:

  • Forms, not wiring — each model exposes only the controls it actually
    understands.
  • Accounts, groups, admin panel — one GPU, many users, per-user presets
    and model access.
  • Drive it from Claude Desktop or any MCP client — per-user tokens; every
    write action needs your approval.
  • Video Director — compose shots in sections instead of one giant prompt.

Beta 0.2.0 · Linux and Windows x86_64 + NVIDIA · WSL2 or
Docker ·
Discord · Reddit ·
Ko-fi

60 seconds to first image

PotionUI is a self-hosted app for generating images, video and music with your own GPU, built to be shared: every person gets their own login, presets and limits. Pick a model, type, and watch it render live.

git clone https://github.com/PotionUI/PotionUI.git potionui && cd potionui
./potionui doctor    # check prerequisites, with a repair command for anything missing
./potionui start     # create the venv, install deps, launch backend + frontend, print the URL

Native Windows: .\potionui.cmd doctor and .\potionui.cmd start from PowerShell or cmd.

Floor: 8 GB VRAM + 16 GB RAM (SDXL). More in Install.

What you can do

The generate workspace with the Krea-2 render Generate images
Pick a model and get a form with only the controls it understands. Previews stream in while it renders, and every tab keeps its own preset, prompts and results.
Generating images
The image editor on the Looks tool, with a look tray, the edit stack and a before and after toggle Edit images
Edit with the Qwen-Image and FLUX 3 edit modes, or fix a picture in the built-in editor: adjust, crop, paint and layer, then save a new copy.
Image editing
An LTX-2.5 video rendering, with a live preview, step progress and the form beside it Generate video
Text to video and image to video with Wan, LTX and MiniMax-H3 (which also takes reference images), with audio where the model makes it. Each model's modes sit behind one picker.
Video models
The video editor trimming a clip, with the timeline, the edit stack and a film grain edit Edit video
Restore, colorize, relight, extend, remove objects, refine faces and upscale a clip, keeping the source's audio and length.
LTX modes
The Video Director with three chained shots Video Director
Compose a video from global, timed and chained sections instead of one giant prompt, each shot starting from the last frame of the one before.
Video Director
The phrasebook with a preview image for every value Prompt tools
Save segments once and reuse them, autocomplete known terms from the phrasebook, and vary prompts with {a|b}, weights and ${vars}.
Prompts workspace
History with collections, search and filters History and Library
Everything you generate is saved with the exact parameters that made it. Tag, search, sort into collections, compare two results, and run media tools on the Library.
History and tags
Admin preset management with an overview of one preset Administration
Users, groups and access, presets and models per user, backends, plans, recipes and backups, all from one panel.
Administration
The assistant proposing a rewrite for each prompt segment, each with an Apply button PotionAI and MCP
PotionAI brainstorms and rewrites prompts and adjusts the form, with your approval on every change. Per-user MCP tokens let Claude Desktop or any MCP client drive your instance.
MCP
The plugins page with the ComfyUI import designing a preset form from a workflow Plugins and presets
Each preset targets one model and renders its own form. Plugins add providers, backends, pipes, field types and cloud models, and your ComfyUI workflows import as presets.
Plugins

Models

Kind Families
Image SDXL · Flux 1 / 2 Klein · FLUX 3 Image Klein · Qwen-Image · Krea-2 · Z-Image · Anima · MOSS-Image · MiniMax-H3-Image
Video Wan 2.1 / 2.2 · LTX-2 / 2.3 / 2.5 · MiniMax-H3 · RIFE (frame interpolation)
Music MiniMax-Music3 · YuE2
3D TRELLIS.2 · Pixal3D
Upscale and restore SeedVR2 (image and video)

No model weights ship with PotionUI. Models download at your request, each under its own license. See Models and licensing.

Install

PotionUI runs on Linux and Windows x86_64 with an NVIDIA GPU.

Way in How
Linux ./potionui doctor, then ./potionui start
Windows .\potionui.cmd doctor, then .\potionui.cmd start from PowerShell or cmd
Docker docker run --gpus all -p 26730:26730 ghcr.io/potionui/potionui:latest (details)
Pinokio Add the launcher to Pinokio from its address, https://github.com/PotionUI/potionui-pinokio, then press Install and Start
Requirement Needed
GPU NVIDIA, Turing or newer, driver 580 or newer
Memory 8 GB VRAM + 16 GB RAM to run SDXL; larger families need more (hardware requirements)
Python 3.12
Node.js 22 or newer (24 LTS recommended)

[!NOTE]
PotionUI is in beta: expect rough edges between releases, and back up anything you care about.

No GPU here? ./potionui start --profile remote dispatches to a worker. Profiles, Windows prerequisites, WSL2, running the backend and frontend separately and logs are in the install guide.

Documentation

Changelog

See CHANGELOG.md, or the GitHub releases page.

Contributing

  • CONTRIBUTING.md: dev setup, test and lint commands, PR expectations.
  • CLAUDE.md: the architecture reference; package-layering rules are enforced by tests/architecture/test_layering.py.
  • Security issue? See SECURITY.md.

Support

If PotionUI is useful to you, a Ko-fi contribution helps keep it going. Development happens in the open on Discord; news, showcases and questions also go to r/PotionUI.

License

PotionUI is GPL-3.0: see LICENSE. Third-party code is bundled under vendor/, each component keeping its own upstream license; full attribution in vendor/NOTICE.md.


Docs ·
Discord ·
Contributing

Reviews (0)

No results found