Install ComfyUI Custom Nodes Without Breaking ComfyUI
Install ComfyUI custom nodes with Manager or manual Git clone, avoid dependency drift, and fix red nodes, import failures, or Manager node-list errors safely.
Start with Wonderful Launcher if this issue touches your real ComfyUI environment. Use the docs to understand the fix, and use the app to inspect the machine you already have.
Download Wonderful LauncherTested Environment
- OS: Windows 10 / 11
- Launcher: Wonderful Launcher v1.x
- ComfyUI: Portable / Managed install
- Python: 3.11+
- CUDA / Torch: CUDA 12.x / Torch 2.x
- Last tested: 2026-05-19
If you want to install ComfyUI custom nodes, use Manager for common nodes and manual Git clone only when you know the repo you need.
Custom nodes are community-built plugins that add new functionality to ComfyUI, from face restoration to video generation to advanced upscaling. They are also one of the fastest ways to break an otherwise working environment.
That usually happens because the install is treated like a simple copy operation, when in reality each custom node may bring new Python packages, binary dependencies, version conflicts, or imports that only fail after restart.
If your environment already broke
If you already installed custom nodes and ComfyUI stopped working, go straight to Installed Custom Nodes and Broke ComfyUI? Recover Without Reinstalling.
Fast Answer
| Goal | Best First Step |
|---|---|
| Install a common custom node | Use ComfyUI Manager from the toolbar |
| Install a node from GitHub manually | Clone it into ComfyUI/custom_nodes/, then install only that node's requirements |
| Fix red nodes in an imported workflow | Use Manager's missing-node install flow, then restart and check IMPORT FAILED |
| Avoid breaking Torch, NumPy, or OpenCV | Inspect requirements.txt before running broad package installs |
Manager says failed to get custom node list | Use the dedicated failed custom node list guide |
If your search was specifically ComfyUI Manager failed to get custom node list, this page is background. The repair path is ComfyUI Failed to Get Custom Node List.
Installing Custom Nodes
Method 1: ComfyUI Manager (Recommended)
ComfyUI Manager is the official package-management layer for recent ComfyUI builds. On current Desktop builds it is already included. On recent portable or manual installs it is built into core, but may need its dependencies installed and the manager flag enabled.
Desktop users
ComfyUI Desktop already ships with Manager enabled by default. Restart the app and look for the Manager button in the toolbar.
Recent portable users
From the portable package root, install manager dependencies and launch ComfyUI with Manager enabled:
.\python_embeded\python.exe -m pip install -r .\ComfyUI\manager_requirements.txt
.\python_embeded\python.exe -s .\ComfyUI\main.py --windows-standalone-build --enable-managerRecent manual-install users
Activate the same venv you use to launch ComfyUI, then run:
python -m pip install -r manager_requirements.txt
python main.py --enable-managerLegacy fallback
If your build still expects the older plugin-style install, use the legacy clone method documented by the ComfyUI Manager team:
cd ComfyUI/custom_nodes
git clone https://github.com/ltdrdata/ComfyUI-Manager.git comfyui-managerOnce Manager is available, use the Manager button in the ComfyUI toolbar to:
- Browse available custom nodes
- Install nodes with one click
- Update installed nodes
- Install missing nodes from imported workflows
Method 2: Manual Git Clone
- Open a terminal in
ComfyUI/custom_nodes/ - Clone the node repository:
cd ComfyUI/custom_nodes
git clone https://github.com/[author]/[node-name].git- Install dependencies only if the node has a
requirements.txt, and use the same Python environment that starts ComfyUI:
| Install type | Safer dependency command |
|---|---|
| Official GitHub Windows portable package | From the portable package root, run .\python_embeded\python.exe -s -m pip install -r .\ComfyUI\custom_nodes\[node-name]\requirements.txt |
| Manual Git + venv install | Activate the venv, then run python -m pip install -r custom_nodes/[node-name]/requirements.txt |
| ComfyUI Desktop or managed launcher | Use the app's environment/terminal tools. Do not assume python_embeded exists. |
- Restart ComfyUI
Do not blindly run requirements on every node
The highest-risk step is usually pip install -r requirements.txt. Before running it, check whether the plugin is about to change Torch, numpy, opencv-python, pillow, or other core packages.
Method 3: Copy Files
Some simple nodes are just a single .py file. Copy it directly into custom_nodes/:
ComfyUI/
+-- custom_nodes/
+-- my_node.pyRestart ComfyUI to load it.
Popular Custom Nodes
| Node | What It Does |
|---|---|
| ComfyUI-Manager | Browse and install other custom nodes |
| ComfyUI-GGUF | Load quantized GGUF models for lower VRAM usage. See ComfyUI GGUF |
| ComfyUI-Impact-Pack | Face detection, segmentation, detailing |
| ComfyUI-AnimateDiff | Animate images into short videos |
| ComfyUI-IPAdapter | Use reference images to guide generation style |
| ComfyUI-ControlNet | Advanced ControlNet support (pose, depth, canny) |
| ComfyUI-VideoHelperSuite | Load, preview, and save video files |
| ComfyUI-KJNodes | Collection of useful utility nodes |
| rgthree-comfy | Workflow organization and quality-of-life nodes |
| efficiency-nodes-comfyui | Batch processing and workflow efficiency tools |
Fixing "Red Nodes"
Red nodes in a workflow mean ComfyUI cannot find a backend node class required by that workflow. That is related to plugins, but it is not the same thing as a successful plugin install.
The practical chain is:
- The workflow asks for a node class.
- A custom node plugin must register that class during startup.
- The plugin can only register if its Python imports succeed.
- Those imports may depend on packages, native wheels, CUDA, PyTorch, or platform-specific binaries.
So a red node can mean "plugin not installed", but it can also mean "plugin folder exists, but the plugin failed to import".
With ComfyUI Manager:
- Click the Manager button
- Click Install Missing Custom Nodes
- The manager will identify and install what's needed
- Restart ComfyUI
Without Manager:
- Read the node name shown on the red node
- Search for it on GitHub
- Install manually via git clone
- Restart and check the startup log for
IMPORT FAILED
Updating Custom Nodes
With ComfyUI Manager: Click Manager > Update All
Manually:
cd ComfyUI/custom_nodes/[node-name]
git pull
python -m pip install -r requirements.txt # only if this is the Python that starts ComfyUIFor the official Windows portable package, run dependency installs from the portable package root with .\python_embeded\python.exe -s -m pip ... instead of relying on a random system pip.
Restart ComfyUI after updating.
Install custom nodes with less risk
If this machine matters, follow this order:
- Install one node or one node family at a time
- Restart ComfyUI after each change
- Check the startup log for
IMPORT FAILED - Avoid stacking multiple
requirements.txtinstalls without testing - Stop as soon as you see dependency drift
That approach is slower at first, but much faster than repairing a broken environment later.
Troubleshooting
| Problem | Solution |
|---|---|
| Red nodes after importing a workflow | Install missing nodes via Manager or manually |
| Node installed but not showing up | Restart ComfyUI. Check the console for import errors |
ModuleNotFoundError on startup | The plugin imported a package that is not in ComfyUI's Python. Install it in the exact environment that starts ComfyUI, or use the plugin's documented wheel. |
| Node conflicts / crashes | Some nodes are incompatible with each other. Disable by renaming the folder (e.g., add .disabled). See Dependency Conflicts |
| Manager cannot load the custom node list | Check GitHub/raw GitHub access, Manager cache, and proxy rules. See ComfyUI Failed to Get Custom Node List |
Going Deeper
For advanced plugin management, node mapping, repo migrations, batch installation, and Git LFS issues, see the Plugin Management guide. If you're setting up a complex multi-workflow environment, see Workflow Environment Setup.
For systematic error diagnosis, use the Troubleshooting Decision Tree.
If you want a more recovery-oriented path than repeated manual installs, start with Wonderful Launcher. If the setup is already unstable, use expert help.
FAQ
What is the safest way to install ComfyUI custom nodes?
Use ComfyUI Manager when it is available. For manual installs, clone the node into ComfyUI/custom_nodes/, install only that node's requirements in the same Python environment that starts ComfyUI, and restart before testing.
Why did installing a custom node break ComfyUI?
The node may have changed shared Python packages such as Torch, NumPy, OpenCV, Pillow, Transformers, or Diffusers. If ComfyUI fails after the install, read the first IMPORT FAILED or traceback before running more package commands.
What should I do if Manager cannot get the custom node list?
That is usually a Manager registry, network, cache, or GitHub access problem, not a normal custom-node installation step. Use ComfyUI Failed to Get Custom Node List.
Related Guides
- ComfyUI Failed to Get Custom Node List
- Installed Custom Nodes and Broke ComfyUI
- How to Fix ComfyUI Plugin Import Failed Errors
- ComfyUI Dependency Conflicts
Source References
Start with Wonderful Launcher if this issue touches your real ComfyUI environment. Use the docs to understand the fix, and use the app to inspect the machine you already have.
Download Wonderful LauncherStart with Wonderful Launcher if this issue touches your real ComfyUI environment. Use the docs to understand the fix, and use the app to inspect the machine you already have.
Download Wonderful LauncherDid this fix your issue?
Your answer helps prioritize verified ComfyUI repairs.
OpenCV / cv2 Missing
Fix No module named 'cv2' in ComfyUI, missing opencv-python-headless, opencv-contrib-python, and OpenCV package conflicts.
Plugin Import Failed
Fix ComfyUI plugin import failed errors, missing custom nodes, red workflow nodes, and dependency drift without reinstalling your environment.