Easily manage multiple ComfyUI installations with different versions, Python setups, and GPU configurations - all from one simple script.
ComfyEnv automatically creates a centralized resource management system that keeps all your files organized and accessible across every environment:
- One central location for all your models - Every installation automatically connects to a single
models/folder at the root level - No duplicate downloads - Download each model once and it's available across all your setups
- Saves disk space - No need to store multiple copies of the same checkpoints, LoRAs, VAEs, etc.
- Always accessible - Your models stay with you even when you create or delete installations
- Centralized workflow storage - All your
.jsonand.pngworkflow files live in oneworkflows/folder - Access from any environment - Load and save workflows from any ComfyUI version seamlessly
- Automatic symlink creation - Each environment automatically links to the shared workflows folder
- Cloning support - When you clone an environment, workflow symlinks are preserved
- Easy sharing & backup - Share your entire workflow library by sharing one folder!
- Centralized file I/O - Single
inputs/andoutputs/folders for all environments - No more scattered files - All your input images and generated outputs in predictable locations
- Automatic organization - Each environment gets its own subfolder within shared spaces
- Easy file management - Find and organize your assets without digging through multiple directories
- Import your ComfyUI settings once - Place your
comfy.settings.jsonin the rootconfigs/folder - Find your config file: Located at
comfyui/user/default/comfy.settings.jsonin any existing ComfyUI installation or inside an environment (./environments/[env-name]/comfyui/user/default/) - Consistent experience - All new environments are created with your settings and preferences
- Global workflow library - Create
workflows/_shared/to store workflows accessible from ALL environments - Automatic linking - Appears as a
shared/subfolder inside each environment's workflow folder - Perfect for: Frequently-used workflows you want everywhere
- Install any ComfyUI version - Supports all versions from v0.3.x to latest
- Automatic Python setup - Smart selection based on what your version needs
- Custom names for your setups - Give them meaningful names like
production.v1ordev-test - Browse available versions - See all GitHub releases
- Run multiple instances simultaneously - Start different ComfyUI versions at the same time on different ports
- Automatic port management - Auto-selects available ports starting from 8188 when default is in use
- Custom port option - Use
--port PORTto specify a specific port number - Instance tracking - Built-in system tracks all running instances with PID files
- Smart cleanup - Automatically handles crashed sessions and stale processes
- Auto-detects your graphics card - Works with NVIDIA, AMD, Intel Arc, and Apple Silicon
- Stable or experimental options - Choose between reliable or cutting-edge performance
- Follows official ComfyUI documentation installation commands
| Command | Description |
|---|---|
--env-list |
See all your installed setups with version info |
--env-delete NAME |
Safely remove a setup (asks for confirmation first) |
--env-rename OLD NEW |
Rename or move a setup directory |
--env-clone SRC DST |
Duplicate an existing setup quickly |
- ComfyUI-Manager support - Installed automatically by default
- Detailed progress output - See exactly what's happening during installation
- Flexible naming - Use letters, numbers, underscores, hyphens, and dots in names
- Symlink repair tool -
--fix-symlinksoption to automatically detect and fix broken symlinks
For detailed information about recent changes, new features, and improvements, please see our CHANGELOG.md file.
Latest Updates (2026-04-17):
- π§ New
--fix-symlinksoption to automatically repair broken symlinks - π― Enhanced rename functionality now handles workflows, inputs, and outputs folders
- π Centralized I/O management with symlink support for input/output folders
- π Workflow symlink management with cloning support
- π
Release dates displayed in
--listoutput - π·οΈ Environment creation and last edit dates shown in lists and menus
- π More reliable version detection using
comfyui_version.py - π Multi-instance support - run multiple ComfyUI instances simultaneously on different ports
- π Shared
_sharedworkflows directory for globally accessible workflow library - βοΈ Custom
comfy.settings.jsonauto-applied fromconfigs/folder during installation
# On Fedora/RHEL/CentOS:
sudo dnf install git curl python3
# On Ubuntu/Debian:
sudo apt install git curl python3
# On Arch Linux:
sudo pacman -S git curl pythonUsing pyenv-installer:
curl https://pyenv.run | bashManual installation:
git clone https://github.com/pyenv/pyenv.git ~/.pyenv
# Add to ~/.bashrc or ~/.zshrc:
export PYENV_ROOT="$HOME/.pyenv"
export PATH="$PYENV_ROOT/bin:$PATH"
eval "$(pyenv init -)"Verify installation:
pyenv --version # Should output version number./install_comfy.sh v0.3.62./install_comfy.sh v0.18.0 --name my-production-env./install_comfy.sh --list# Install specific version with auto-detected Python
./install_comfy.sh v0.3.62
# Force specific Python version
./install_comfy.sh v0.18.0 --python 3.13
# Skip ComfyUI-Manager installation
./install_comfy.sh v0.3.62 --no-manager# Create environment with custom name
./install_comfy.sh v0.3.62 --name production.v1
# Valid characters: a-z, A-Z, 0-9, _, -, .
./install_comfy.sh v0.18.0 --name dev-test-env./install_comfy.sh --env-list
# Output:
# NAME VERSION PYTHON PATH
# ---- ------- ------ ----
# v0.3.62 v0.3.62 python_3.11 /path/to/environments/v0.3.62/
# production.v1 v0.18.0 python_3.13 /path/to/environments/production.v1/./install_comfy.sh --env-delete old-environment-name./install_comfy.sh --env-rename old-name new-name# Creates a complete copy (faster than fresh install)
./install_comfy.sh --env-clone source-env target-env./install_comfy.sh --helpAfter installation, your folder will look like this:
comfy/
βββ install_comfy.sh # Main installer script
βββ models/ # β Shared models - all environments use this!
β βββ checkpoints/
β βββ vae/
β βββ loras/
βββ workflows/ # π Shared workflows - accessible from all environments
β βββ my-environment/
β β βββ my-workflow1.json
β β βββ my-workflow2.json
β βββ _shared/ # π Global workflow library (linked as 'shared/' in each env)
β βββ utility-workflow.json
βββ inputs/ # π₯ Shared inputs folder
β βββ v0.3.62/ # Per-environment subfolders
β βββ production.v1/
βββ outputs/ # π€ Shared outputs folder
β βββ v0.3.62/
β βββ production.v1/
βββ configs/ # βοΈ Initial configuration folder
β βββ comfy.settings.json # Your ComfyUI settings (imported to all envs)
βββ environments/ # All installations stored here
βββ v0.3.62/
β βββ python_3.11/ # Python setup
β βββ comfyui/ # ComfyUI source code
β βββ models -> ../../models/ # β Shared models
β βββ workflows -> ../../workflows/ # β Shared workflows
β βββ inputs -> ../../inputs/v0.3.62/# β Per-env input folder
β βββ outputs -> ../../outputs/v0.3.62/# β Per-env output folder
βββ production.v1/
βββ python_3.13/
βββ comfyui/
βββ models -> ../../models/ # β Shared models
βββ workflows -> ../../workflows/ # β Shared workflows
βββ inputs -> ../../inputs/production.v1/# β Per-env input folder
βββ outputs -> ../../outputs/production.v1/# β Per-env output folder
./install_comfy.sh v0.3.62
# Select: 1) NVIDIA (CUDA)
# Choose stable or nightly PyTorch when promptedPyTorch versions:
- Stable: CUDA 13.0 - Production-ready
- Nightly: CUDA 13.2 - Latest features, may have better performance
./install_comfy.sh v0.3.62
# Select: 2) AMD (ROCm)Note: Ensure ROCm drivers are installed on your system.
./install_comfy.sh v0.3.62
# Select: 3) Intel (XPU)./install_comfy.sh v0.3.62
# Select: 4) Apple SiliconNote: Uses nightly PyTorch for best Metal performance.
# Add to ~/.bashrc or ~/.zshrc:
export PYENV_ROOT="$HOME/.pyenv"
export PATH="$PYENV_ROOT/bin:$PATH"
eval "$(pyenv init -)"
# Reload shell:
source ~/.bashrc # or source ~/.zshrc# List available versions:
pyenv install --list | grep "^ 3."
# Install specific version:
pyenv install 3.13.0- NVIDIA: Ensure CUDA drivers are up to date
- AMD: Verify ROCm is properly installed
- Try switching between stable/nightly PyTorch versions
# List existing environments:
./install_comfy.sh --env-list
# Delete if needed:
./install_comfy.sh --env-delete conflicting-name- ComfyUI Official Repository - Main project
- PyTorch Installation Guide - GPU-specific builds
- ComfyUI Documentation - Examples and tutorials
- Pyenv Documentation - Python version manager
This project was vibecoded with: Qwen3.5-27B-Claude-4.6-Opus-Reasoning-Distilled-v2-GGUF
If this project helped you, consider buying me a coffee!