WAS Node Suite for ComfyUI (2026): Install v3, Best Nodes, Old Workflows and Fixes

WAS Node Suite is one of the oldest and biggest custom node packs for ComfyUI. If you have ever opened a shared workflow and seen nodes like Image Save, Text Concatenate, Image Blending Mode or Logic Boolean, they almost certainly came from WAS.

The pack changed a lot recently. Version 3 is now out with almost 700 nodes, no automatic pip installs and a new warning about ComfyUI Manager. There is also a separate “Revised” fork that many people installed while the original author was away.

This guide explains what WAS Node Suite is, which version to install, how to install it safely, the most useful nodes, how to open old workflows and how to fix the common errors.

What WAS Node Suite Is

WAS Node Suite (often shortened to WAS-NS) is a custom node pack for ComfyUI made by Jordan Thompson, known online as WASasquatch. It has been around since 2023 and has over a million downloads.

Instead of doing one job, it adds a huge toolbox of small nodes. You use them to fill gaps in ComfyUI’s built-in nodes, such as:

• Image editing: blending modes, filters, film grain, colour matching, levels, LUTs, cropping and padding.
• Masks: combine, grow, erode, blur and crop masks.
• Text and prompts: join strings, find and replace, load text files, wildcards and prompt libraries.
• Logic and numbers: booleans, comparisons, switches, maths and random numbers.
• Files: save images with custom folders and names, load image batches from a folder, save text and documents.
• Video and animation: frame tools, interpolation and video saving.
• LLM tools: a Llama-Index group that brings document loaders, retrievers, chat engines and agents onto the canvas.

It is MIT licensed, so you can use it in personal and commercial work, including paid services.

v3 vs v2 vs the Revised Fork

There are now three things people call “WAS Node Suite”. Knowing which one you have avoids most problems.

	WAS Node Suite v3	WAS Node Suite v2 (old)	WAS Node Suite (Revised)
Repository	WASasquatch/was-node-suite-comfyui	Same repo, older releases	ltdrdata/was-node-suite-comfyui
Nodes	About 695	About 220	The v2 set, maintained
Packages installed by default	0	About 20	Similar to v2
Downloads at runtime	Off by default	Some nodes downloaded models	Some nodes download models
Status	Current, active	Superseded	Registered in June 2025 when the original author retired

The Revised fork was registered by ltdrdata (the ComfyUI Manager author) on 3 June 2025 after WASasquatch announced their retirement, so existing workflows kept working. WASasquatch has since returned with v3 in the original repository.

Which should you use? For a new setup, use v3 from the original repository. If you have an old, working setup with the Revised fork and do not want any changes, you can stay on it, but do not install both packs at the same time.

How to Install WAS Node Suite v3

Before you start, make sure ComfyUI is up to date. v3 needs ComfyUI 0.14.0 or newer and Python 3.10 or newer. If you need help, see how to update ComfyUI.

Method 1: Git clone (recommended by the author)

1. Close ComfyUI if it is running.
2. Open a terminal in the custom_nodes folder. For the Windows portable build this is ComfyUI_windows_portable\ComfyUI\custom_nodes. For a manual install it is ComfyUI/custom_nodes.
3. Clone the repository:

   git clone https://github.com/WASasquatch/was-node-suite-comfyui.git

4. Start ComfyUI. The first start takes a second or two longer while the pack writes its config and compiles itself.
5. Check it loaded. Double-click the canvas and search for “WAS” or a node such as “Image Save”. The nodes should appear.

That is the whole install. v3 does not run pip and does not download anything on its own, now or on later starts.

Method 2: Open Manager

The author also points to their own Open Manager extension (WASasquatch/open-manager-comfyui), which can install v3 and its optional groups from an Optional tab.

What about ComfyUI Manager?

The v3 README carries a clear warning: do not install v3 through ComfyUI Manager unless you are installing an approved version, because it can silently downgrade your install and break things. If you normally use ComfyUI Manager, use the git clone method for WAS and keep Manager for other packs.

Desktop app users

In the Comfy Desktop app, your installation still has a custom_nodes folder inside its ComfyUI install directory. Open that folder in a terminal and use the same git clone command, then restart the app.

Optional Feature Groups (LLM, Documents)

To keep the base pack light, four groups need extra Python packages and ship turned off:

Group	What it adds	Config key
document_export	Save DOC can write .docx, .odt and .pdf	document_export: true
llm	The language model and Llama-Index nodes	llm: true
vector_stores	LanceDB, FAISS and Postgres vector stores for the LLM nodes	vector_stores: true
chroma	Chroma vector store	chroma: true

To turn one on with the Windows portable build, run this from the ComfyUI_windows_portable folder (example for the LLM group):

python_embeded\python.exe -m pip install -r ComfyUI\custom_nodes\was-node-suite-comfyui\requirements\llm.txt

Swap llm.txt for document_export.txt, vector_stores.txt or chroma.txt as needed. On a manual install, activate your ComfyUI virtual environment and run pip install -r with the same file.

Then open the pack’s config.yaml and set the matching key to true under features:. Restart ComfyUI.

The LLM nodes can use models your graph already loads, GGUF files, Hugging Face folders, local servers such as Ollama, or hosted providers. If you want a simpler way to add text generation to a workflow, see our guide to generating text with an LLM in ComfyUI.

Where the Config File Lives

On first start, v3 writes its settings to <ComfyUI user dir>/was-node-suite/config.yaml. For most installs the user directory is ComfyUI/user, so the file is at ComfyUI/user/was-node-suite/config.yaml. The same folder holds its state database and the wildcard, LUT and view-extension folders.

Edit the config there, not inside the pack folder, because this copy survives updates. Useful keys include:

• features: the optional groups above, plus network.
• network: false by default. Set it to true only if you want nodes that need model weights to download them automatically.
• legacy groups: 28 deprecated nodes sit in nine legacy groups, four of which start off.

Older v2 installs used a was_suite_config.json file inside the pack folder instead. If you see that file, you are on v2 or the Revised fork.

The Most Useful WAS Nodes

With nearly 700 nodes it is easy to get lost. These are the ones most people use every day.

Node	What it does	Typical use
Image Save	Saves images with custom folder, prefix, format and overwrite mode	Organised output folders per project
Load Image Batch	Loads images from a folder one by one or all at once	Batch upscaling or captioning
Text Concatenate	Joins several text strings	Build prompts from parts
Text Find and Replace	Replaces words in a string	Swap a subject across prompts
Image Blending Mode	26 blend modes, computed in linear light on the GPU	Overlay, screen, multiply effects
Image Film Grain	Adds brightness-aware film grain	Make AI images look more photographic
Image Style Filter	37 colour-graded looks	Quick film or vintage looks
Logic Boolean / Logic Compare Numbers	True or false values and comparisons	Switch parts of a workflow on and off
Random Number / Seed (Number Outputs)	Generates numbers and seeds	Share one seed across several samplers
Lora Loader (Advanced)	Loads a LoRA with model and CLIP strengths	Replaces the old WAS Load Lora
Mask nodes (Combine, Erode Region, and more)	Edit and combine masks	Inpainting prep

The full list with every input and output is in NODES.md in the repository. In ComfyUI, you can also double-click the canvas and type a word like “mask” or “text” to see all matching WAS nodes.

Example: Save Images With Custom Names

A common reason to install WAS is better file saving. Here is a simple setup:

1. Open any text-to-image workflow, for example a template from our ComfyUI workflows guide.
2. Delete or bypass the built-in Save Image node.
3. Add WAS Image Save and connect the VAE Decode IMAGE output to it.
4. Set output_path to the folder you want, for example a project name such as portraits.
5. Set filename_prefix, the file format (PNG, JPG or WEBP) and quality.
6. Leave overwrite_mode off unless you want each run to replace the same file.
7. Run the workflow and check your output folder.

Hover over each widget in ComfyUI to see its tooltip. NODES.md in the repository lists every input with its description.

Opening Workflows Made Before v3

The author kept old workflows working where possible. 213 of the 220 v2 node IDs are unchanged, and widget values are carried across when a workflow loads.

Four nodes are retired and replaced. They show up in ComfyUI’s missing-node dialog. Tick them and press the replace button.

Saved as	Comes back as
Image Crop Face	Same name, new code using the same eight face cascades
Image Gradient Map	Same name, new code
Load Lora	Lora Loader (Advanced)
Number to Text	Number to String

Five nodes were removed with no replacement, mostly because they made network requests: Image Send HTTP, Text Random Prompt, True Random.org Number Generator, WASDownloadImage and WASNoodleSoupPantryRefresh. Delete them and use Image Save, a written prompt or Prompt Library, Random Number or Seed, and Load Image instead.

Three labels were renamed so they do not clash with core nodes: KSampler is now KSampler (Seed Socket), Seed is Seed (Number Outputs) and Save Video is Save Video (Advanced). Search still finds them by the old names.

Watch for changed results. Image Blending Mode’s add mode now really adds the layers, so old graphs using it come out brighter (switch to normal for the old look). Logic Boolean now reads any non-zero number as true. Image SSAO’s depth input is now height_maps and must be re-wired.

Updating and Uninstalling

To update v3: open a terminal in custom_nodes/was-node-suite-comfyui and run:

git pull

Restart ComfyUI. The first start after an update is slightly slower because the pack recompiles.

To switch from the Revised fork to v3: close ComfyUI, move the old was-node-suite-comfyui folder out of custom_nodes (keep it as a backup), then clone v3 as shown above.

To uninstall: close ComfyUI and delete the custom_nodes/was-node-suite-comfyui folder. If you also want to remove its settings, delete ComfyUI/user/was-node-suite. Workflows that use WAS nodes will then show those nodes as missing.

Troubleshooting

Problem	Likely cause	Fix
WAS nodes do not appear	Wrong folder or ComfyUI too old	Check the pack is in ComfyUI/custom_nodes, update ComfyUI to 0.14.0+, read the startup log for “IMPORT FAILED”
Pack downgraded after using Manager	Manager installed an older version	Delete the folder and reinstall with git clone
Red “missing node” boxes in an old workflow	Retired or removed v2 nodes	Use the replace button in the missing-node dialog, or delete the five removed nodes
Duplicate or conflicting nodes	Both v3 and the Revised fork installed	Keep only one was-node-suite-comfyui folder
A node says it cannot find model weights	network is false	Place the weights as described in docs/MODELS.md, or set network: true
LLM or Save DOC nodes missing	Optional group not installed or not enabled	Install the matching requirements file and set its key to true in config.yaml
Image looks brighter than before	Image Blending Mode add changed in v3	Switch the mode to normal
git is not recognised	Git not installed	Install Git for Windows, or download the repo ZIP and extract it into custom_nodes

If ComfyUI itself will not start after adding nodes, temporarily move the new folder out of custom_nodes to confirm it is the cause. Our ComfyUI beginners guide covers reading the startup log.

My Take

WAS v3 is a big clean-up. The old version installed about 20 packages and made web requests from some nodes, which caused dependency fights with other packs. v3 installs nothing by default and keeps network access off, which is safer and much friendlier to busy ComfyUI setups.

For a new install, clone v3 from the original repository and skip ComfyUI Manager for this one pack. If an old workflow breaks, the missing-node dialog fixes most issues in a click.

FAQ

Is WAS Node Suite free?

Yes. It is MIT licensed, so you can use it for personal and commercial projects.

Is WAS Node Suite still maintained?

Yes. The original author retired in mid-2025, and ltdrdata registered a Revised fork to keep it working. The author has since released v3 in the original repository, which is the actively developed version.

Should I install WAS through ComfyUI Manager?

The v3 README warns against it unless it is an approved version, because Manager can silently downgrade the pack. Use git clone or Open Manager.

Does WAS Node Suite slow down ComfyUI?

Only the first start after install or update is a second or two slower while it compiles. v3 installs no packages by default, so it does not add heavy dependencies.

Where is the WAS config file?

For v3 it is ComfyUI/user/was-node-suite/config.yaml. Older versions used was_suite_config.json inside the pack folder.

Will my old workflows still work?

Mostly yes. 213 of 220 old node IDs are unchanged. Four retired nodes can be replaced from the missing-node dialog, and five removed nodes must be deleted.

Can I use WAS Node Suite on Comfy Cloud?

Custom node packs you install yourself work on local ComfyUI. On cloud services, only the packs the provider supports are available, so check their list.

Sources

• WASasquatch/was-node-suite-comfyui (v3 README, NODES.md, docs/CONFIG.md)
• ltdrdata/was-node-suite-comfyui (Revised fork)
• WASasquatch/open-manager-comfyui
• ComfyUI docs: How to update ComfyUI