ComfyUI Missing Nodes: Fix Red Workflow Nodes and Node Packs

If you searched for ComfyUI missing nodes, missing node packs, or red workflow nodes, this page is for workflow nodes that are red, unavailable, or unresolved after import. First decide which layer is failing: a shared workflow can need a custom node pack, a newer built-in node, a working frontend package, or model files in the right folder.

Use a narrower guide when the main visible error is not a missing node:

• Frontend entry-point message: use the ComfyUI clean library entry point guide.
• ControlNetApplyAdvanced class_type: use the ControlNetApplyAdvanced class_type guide.
• failed to save workflow draft: use Failed to Save Workflow Draft.

This missing-node repair path is for the common case where a workflow stores node type names, but the actual node code is not installed or did not load.

Wonderful Launcher telemetry for the 30 days ending June 29, 2026 showed missing workflow assets as a real repeated pattern: missing models were detected on 27 installs, missing nodes on 19 installs, missing-node clone results on 21 installs, and dependency install results on 17 installs. That usually means a workflow repair has stages: install the node, load the plugin, install its dependencies, then place the model files.

What "missing nodes" usually means

In practice, missing nodes usually fall into one of these buckets:

What it looks like

When you load a workflow with missing nodes, ComfyUI shows:

When loading the graph, the following node types were not found:
  - KSamplerAdvanced (Custom)
  - ControlNetApplyAdvanced
  - IPAdapterApply
  - CLIPTextEncodeSDXL

In the workflow canvas, missing nodes appear as red rectangles with the node type name displayed.

If it is a frontend entry-point message

Treat frontend entry-point wording as a workflow/frontend compatibility clue before installing random Python packages. Use the dedicated ComfyUI clean library entry point missing guide if that is the main error on screen. This page stays focused on the broader missing-node workflow path.

Use this order:

1. Update ComfyUI and its frontend packages together if the workflow came from a newer ComfyUI setup.
2. Try opening the workflow in the simplest local URL such as http://127.0.0.1:8188, not a remote/proxy page.
3. Check whether normal built-in nodes still load. If they do not, this is broader than one missing custom node.
4. Look for red custom nodes after the workflow opens. If red nodes remain, continue with missing-node repair.
5. Do not pip install clean-library-entry-point or similar guessed names. That phrase is not a package name.

If the same install also shows a frontend version warning, use comfyui-frontend-package not installed or outdated. If the visible issue is a browser save failure, use Failed to Save Workflow Draft.

If the missing node is ControlNetApplyAdvanced

ControlNetApplyAdvanced is documented by ComfyUI as a built-in ControlNet node. If a workflow complains about ControlNetApplyAdvanced class_type, do not start by running pip install ControlNetApplyAdvanced; that package name is not the repair path.

Use the dedicated ComfyUI ControlNetApplyAdvanced class_type guide if that exact phrase is the main error. Keep this page for broader missing-node triage.

Treat it in this order:

1. Update ComfyUI itself if the workflow came from a newer setup.
2. Check whether the workflow JSON is valid if the message says node id has no class_type.
3. Verify ControlNet model inputs if the node exists but fails when queued.
4. Install ControlNet preprocessors only when the workflow actually uses preprocessor nodes such as Canny, Depth, OpenPose, or similar helper nodes.

This is the same missing-node principle in a more specific shape: first decide whether the node is built in, custom, corrupted in the workflow JSON, or failing at execution time.

Why it happens

• Workflow from another user: They had custom nodes installed that you do not have
• Custom nodes not installed: The workflow uses node packs that must be separately installed
• Custom nodes installed but not loaded: The node pack exists in custom_nodes/ but failed to import
• Different ComfyUI version: Some built-in nodes were added in newer versions
• Node renamed or removed: The custom node pack updated and renamed or removed a node type

The fastest way to avoid wasted installs

Before installing anything, ask three quick questions:

1. Is this a truly custom node or a newer built-in node?
2. Did the workflow come from another machine with extra node packs?
3. Does the startup log show an import failure for a pack you already installed?

That check prevents a lot of blind pip install and duplicate plugin installs.

Step-by-step fix

Step 1: Identify the missing nodes

The popup when loading the workflow lists all missing node types. Write them down or take a screenshot.

If you dismissed the popup, you can also:

• Look for red boxes in the workflow canvas
• Check the ComfyUI startup log for IMPORT FAILED messages
• Open the workflow JSON file and search for node type names

Step 2: Use ComfyUI Manager to find and install them

If you have ComfyUI Manager installed:

1. Click Manager in the ComfyUI menu
2. Click Install Missing Custom Nodes
3. ComfyUI Manager will try to match missing node types to known custom node packs
4. Install the suggested packs
5. Restart ComfyUI

Step 2.5: If "Install Missing Custom Nodes" fails

A failed missing-node install is not the same as a fixed workflow. Before retrying, identify where the install stopped.

Repair stages after importing a workflow

Do not treat the first improvement as the final fix. A shared workflow can fail in several stages:

This staged view is useful because it prevents a common false finish: "red nodes are gone" does not mean the workflow is ready. Queue the workflow once, then follow the newest error.

If a retry gives the same clone or permission error, stop and preserve the current logs. Repeating the same install command usually adds noise without changing the blocker.

Step 3: Manual installation if Manager cannot find them

Some custom nodes are not in ComfyUI Manager's registry. You need to find them manually:

1. Search GitHub for the node type name (e.g., "IPAdapterApply" comfyui custom node)
2. Clone the repo into your custom_nodes/ folder:

cd ComfyUI\custom_nodes
git clone https://github.com/author/ComfyUI-NodePack.git

3. Install the node's requirements:

.\python_embeded\python.exe -s -m pip install -r ComfyUI\custom_nodes\ComfyUI-NodePack\requirements.txt

4. Restart ComfyUI

Step 4: If nodes are installed but still show red

If the custom node folder exists but nodes still show as red or missing:

• Check the startup log for IMPORT FAILED — the node pack may have a dependency error
• The node pack may need to be updated: cd ComfyUI\custom_nodes\<pack> && git pull
• A Python package required by the node may be missing — see the traceback for details

Step 4.5: Separate "missing pack" from "broken pack"

This distinction matters:

• missing pack: the folder or plugin was never installed
• broken pack: the folder exists, but import failed during startup

If the pack exists but the nodes are still red, jump quickly to:

• ComfyUI Plugin Import Failed
• Fix No module named errors in ComfyUI
• ComfyUI Dependency Conflicts

Step 5: Verify

After installing all missing nodes and restarting, load the workflow again. All nodes should appear with their normal colors and be fully functional.

Verification should include the running backend, not only the folder on disk:

• The plugin folder exists under custom_nodes/
• ComfyUI restarted after the install
• Startup logs do not show IMPORT FAILED for that plugin
• The red node types are gone from the workflow
• Any remaining errors are about models, inputs, or execution, not missing node classes

Built-in vs custom nodes

Some nodes that look custom are actually built-in to ComfyUI but were added in a specific version. If a missing node has a generic name (like KSampler or VAEDecode), you may need to update ComfyUI itself:

cd ComfyUI
git pull

When Wonderful Launcher is the safer path

Workflow missing-node repair gets risky when:

• several packs are missing at once
• one install reveals multiple dependency conflicts
• you are not sure whether the node is built-in or third-party
• the workflow matters and you do not want to destabilize a working environment

That is where guided missing-node detection and managed plugin repair usually save more time than manual trial and error.

When not to run pip install blindly

• Do not install requirements from every custom node pack at once — they may conflict
• Read the requirements.txt before running it to check for torch downgrades
• If a node pack is very old or unmaintained, consider finding an alternative
• Some node packs have complex build requirements (C++ compilation, CUDA extensions) that may not work on all systems
• Do not treat a failed Git clone or denied install as a Python dependency problem

How Wonderful Launcher can help

Wonderful Launcher can help detect missing nodes when you import a workflow and guide you through installing the required custom node packs safely. It also helps keep the repair state clear: missing node, install blocked, plugin import failed, or model asset still pending.

Download Wonderful Launcher — it's free and makes workflow sharing easier.

Related errors

• ComfyUI Workflow Red Nodes
• ComfyUI frontend entry-point guide
• ComfyUI ControlNetApplyAdvanced class_type Fix
• Nodes Show Red After Importing Workflow
• ComfyUI Plugin Import Failed
• Install Custom Nodes
• Popular ComfyUI Custom Node Packs
• ComfyUI Failed to Get Custom Node List
• Where to Put Safetensors in ComfyUI
• ComfyUI Manager

Source References

• ComfyUI GitHub
• ComfyUI ControlNetApplyAdvanced built-in node documentation
• Official ComfyUI custom node installation guide
• Official ComfyUI Manager custom node management guide
• ComfyUI Manager GitHub