Save 4 hours! We auto-setup your workflow! Free!

Drop your workflow.json — we handle every dependency, custom node, and model. Just open the link and run.

Auto-Setup Workflow Json (Free) Now!
ComfyUI > Nodes > ComfyUI-GadgetNodes

ComfyUI Extension: ComfyUI-GadgetNodes

Repo Name

ComfyUI-GadgetNodes

Author
2daadv (Account age: 0 days)
Nodes
View all nodes(1)
Latest Updated
2026-11-12
Github Stars
0.0K

How to Install ComfyUI-GadgetNodes

Install this extension via the ComfyUI Manager by searching for ComfyUI-GadgetNodes
  • 1. Click the Manager button in the main menu
  • 2. Select Custom Nodes Manager button
  • 3. Enter ComfyUI-GadgetNodes in the search bar
After installation, click the Restart button to restart ComfyUI. Then, manually refresh your browser to clear the cache and access the updated list of nodes.

Visit ComfyUI Online for ready-to-use ComfyUI environment

  • Free trial available
  • 16GB VRAM to 80GB VRAM GPU machines
  • 400+ preloaded models/nodes
  • Freedom to upload custom models/nodes
  • 200+ ready-to-run workflows
  • 100% private workspace with up to 200GB storage
  • Dedicated Support

Run ComfyUI Online

ComfyUI-GadgetNodes Description

ComfyUI-GadgetNodes adds utility nodes for cleaning and varying prompts, applying model presets, selecting and refining images, and editing training tags in ComfyUI image-generation workflows.

ComfyUI-GadgetNodes Introduction

ComfyUI-GadgetNodes is a collection of utility nodes for ComfyUI. It helps you with the work around image generation: preparing prompts, reusing model settings, organizing images, and editing tags for a training dataset. You can find its nodes under the Gadget/ categories in ComfyUI.

For example, you can turn a rough, multiline prompt into a clean one, separate unwanted details into a negative prompt, choose from saved prompt phrases, or apply your usual settings when you select a checkpoint. Other nodes help you crop or refine a batch of images and manage the text files that accompany training images.

Compatibility note: Some nodes rely on custom controls built for ComfyUI’s traditional LiteGraph interface. Those controls may not work correctly in the newer Nodes 2.0 interface. See ComfyUI-GadgetNodes Troubleshooting before building a workflow around an interactive node.

How ComfyUI-GadgetNodes Works

Each node handles a small task. You connect it to other nodes in a ComfyUI workflow, much like adding a tool to an assembly line:

  1. Prepare what goes in. A prompt node can clean up text, translate it, or select a phrase from a palette.
  2. Choose how to generate. A model node can load a checkpoint or provide settings associated with a selected model.
  3. Work with the images. Image nodes can load a folder, help you choose images, resize them, crop them, or apply finishing steps.

You do not need every node in one workflow. Use the ones that solve a specific problem.

Some features read files that you create and maintain in the ComfyUI-GadgetNodes/ folder. These include model presets in models_config.yaml, tag-editor preferences in train_config.yaml, and prompt collections in prompt/*.yaml. The extension can be used without these files, but the corresponding features will have less information to work with. A YAML file is simply a structured text file for saving settings or lists of prompt phrases.

ComfyUI-GadgetNodes Features

Prompt Writing and Organization

These nodes appear under Gadget / prompt.

NodeWhat it helps you do
Normalize PromptRemoves supported comments (/* */, <!-- -->, #, and //), tidies commas and spaces, and turns line breaks into a single-line prompt. Useful when combining notes from several sources.
Translate PromptTranslates input such as Japanese text into an English prompt. It supports Ollama or Google translation and has a mode for translating only text enclosed in 「…」. In the traditional UI, you can trigger immediate translation with Ctrl+Enter, or Cmd+Enter on Mac.
Analyze PromptChecks a prompt for special markers. ☹ enables its facedetailer_enabled output; ✌ enables handrefiner_enabled. The node removes those markers from the prompt. It also adds explicit and uncensored when it detects certain NSFW-related terms. Review its output if you use it in a workflow.
Split PromptMoves tags written as -(tag) into a negative prompt, leaving the other text in the positive prompt.
Eval PromptsIncludes or excludes prompt text according to conditions, such as the selected model name or words elsewhere in the positive or negative prompt.
Expand WildcardsReplaces wildcard references with entries from text files to produce prompt variations. max_variations limits how many are generated; auto_refresh controls reloading when wildcard files change.
Prompt To FileNameMakes a save-friendly filename from a prompt. You can use %prompt and %time, with time_format controlling the time text.
Prompt PaletteDisplays phrases from prompt/*.yaml in a category tree. Click an item to show its phrase in the text area. The Edit YAML File button opens the file in your system’s default editor.

Separating positive and negative details: If you enter:

1girl, smile, -(lowres), -(bad anatomy:1.2), standing

Split Prompt outputs 1girl, smile, standing as positive text and (lowres), (bad anatomy:1.2) as negative text. The -() notation is a way to keep both kinds of instruction together while you draft.

Saving reusable phrases: You can organize a Prompt Palette file by category. For example, a file at prompt/example.yaml could contain:

Character:
  Casual: "1girl, casual clothes, smile"
  Formal: "1girl, formal dress, elegant pose"
Background:
  Outdoor: "outdoor, sunny day, blue sky"

The category names help you browse; the quoted text is the phrase you select.

Making variations: Expand Wildcards uses the dynamicprompts wildcard format. Put a colors.txt file containing one color per line in ComfyUI/wildcards/, if that folder exists, or in ComfyUI-GadgetNodes/wildcards/. Writing {__colors__} in a prompt lets the node substitute entries from that file. If both wildcard locations are available, ComfyUI/wildcards/ takes priority.

Conditional prompts: Eval Prompts is useful when you share a workflow between models or scenes. For instance, you can include a phrase only when a model name matches a pattern:

<if model="Pony">score_9, score_8_up,</if>
1girl, school uniform

Conditions can check the model name (model), require all listed terms (all), require at least one (any), or match a text pattern (matches). You can combine conditions. The node cleans input and output in the same way as Normalize Prompt, so comments are removed before evaluation. Nested <if> tags are not supported.

Model Settings and Metadata

These nodes appear under Gadget / model. The extension does not provide its own image-generation models. It helps you load models you already use and keep their settings and notes organized.

NodeWhat it helps you do
Load Checkpoint PresetSupplies settings associated with the selected model name, such as a VAE, CLIP settings, steps, sampler, and starting prompts. It uses your models_config.yaml file.
Load Checkpoint or Diffusion ModelProvides one loader for checkpoints and diffusion models. When using a diffusion model, you must also specify clip_name and vae_name.
Edit SD Lora InformationEdits a .json information file beside a LoRA, including its description, activation text, preferred weight, and notes.
Edit SD Checkpoint InformationEdits a .json information file beside a checkpoint or diffusion model, including its description, notes, and recommended VAE.

A preset is a starting set of settings, not a different model. For example, you might give an SDXL-named checkpoint a particular VAE and 30 steps, while a model whose name begins with flux gets a different CLIP setting and 20 steps. Selecting the model then brings the relevant settings into the Load Checkpoint Preset interface.

Here is an abbreviated models_config.yaml example:

__keywords__:
  "{quality}": "masterpiece, best quality"

"SDXL.*":
  steps: 30
  cfg: 7.0
  positive: "{quality}, 1girl, standing"

".*":
  sampler_name: "euler"
  scheduler: "normal"

"SDXL.*" is a name-matching pattern; ".*" is a general fallback. If a name matches more than one pattern, the extension reads them in file order. The first matching pattern wins for any setting it supplies; later patterns fill only settings that are still missing. Place specific preferences before general defaults. The preset_name output lists the patterns that actually supplied values.

The metadata editors keep notes in a .json file beside the model file. For example, my_lora.safetensors can have a neighboring my_lora.json containing an activation phrase and preferred weight. This helps you remember how you intended to use that LoRA; the information file is separate from the model itself.

Image Selection and Finishing

These nodes appear under Gadget / image.

NodeWhat it helps you do
Load Images From FolderLoads multiple images from a folder. You can filter filenames with a regular-expression pattern, limit the number loaded, and include images in subfolders through recursive search.
Select Image IndicesShows images in a preview dialog and outputs the indices you select as comma-separated text. Useful when you want to choose particular images from a list.
Crop Images (Manual)Opens a dialog to set crop areas for images in a list. Aspect-ratio presets include 4:5, 1:1, and 16:9.
Smart Resize ImagesResizes or crops toward a target pixel count and size multiple. If resizing would exceed its distortion threshold, it crops to preserve the image’s proportions.
Refine ImagesCombines a median filter, denoising, Gaussian anti-aliasing, and sharpening for image post-processing.

For a portrait set, you could load a folder, choose the images you want, then use a 4:5 crop preset to prepare them for a consistent layout. For another workflow, Smart Resize Images can help meet a size requirement while avoiding excessive stretching.

Training Tag Editing

Edit Train Tags, under Gadget / train, provides an interface for editing the .txt tag files paired with training images. You can organize keep/remove tags, reorder tags with drag and drop, and preview mask images when present.

The expected pairing looks like this:

dataset/
  image001.png
  image001.txt
  image001-masklabel.png
  image002.jpg
  image002.txt

The .txt files contain comma-separated tags. The mask file is optional. In train_config.yaml, mask_image_postfix identifies mask filenames so they can be excluded from the ordinary image list, while black_list_tag specifies tags treated as blacklisted in the editing interface:

mask_image_postfix: "-masklabel.png"
black_list_tag:
  - "lowres"
  - "bad anatomy"

Workflow Helpers

CategoryNodesWhat they do
Gadget / coreAny Print, Select Input FolderPrint a value to the ComfyUI log for inspection, or select a folder beneath ComfyUI’s input folder and output its path.
Gadget / core / const0, 1, -1, 64, 256, 512, 1024Output commonly used whole-number values without typing them into another node.
Gadget / logicAny Pass, ++1, --1Pass a value through as an any-type value, or increase/decrease a whole number by one.
Gadget / logicInt Equals, Int Not EqualsCheck whether two whole numbers are equal or different; output a true/false result.
Gadget / logicInt Nearly Equals, Float Nearly EqualsCompare numbers approximately. Whole numbers count as nearly equal when they differ by no more than one; floating-point values use an approximate comparison.
Gadget / logicSelect Nth Items Of AnyListPick an item by its position in a list. A negative index counts backward from the end; out-of-range selections are omitted.
Gadget / logicVirtual GroupCollapse or expand canvas nodes by ID range, such as 1-3, 6, or by title with * wildcards. This changes the canvas view, not workflow execution.

ComfyUI-GadgetNodes Troubleshooting

An interactive node does not respond in Nodes 2.0

Some controls were made for ComfyUI’s traditional LiteGraph interface. In Nodes 2.0, the affected behavior depends on the node:

  1. Check whether the issue concerns a custom button, tree, preview, keyboard shortcut, or dialog.
  2. If you can use the traditional interface, try the same node there.
  3. For Crop Images (Manual) and Select Image Indices, do not rely on a run completing in Nodes 2.0 if its dialog never opens.
  4. Test other interactive nodes individually before depending on them in a larger workflow.

Potentially affected nodes are Translate Prompt, Prompt Palette, Load Checkpoint Preset, Edit SD Lora Information, Edit SD Checkpoint Information, Crop Images (Manual), Select Image Indices, Edit Train Tags, and Virtual Group. For Translate Prompt, queued translation may still run even if its shortcut or UI update fails. Nodes without these custom interface features, such as Normalize Prompt and Split Prompt, are generally expected to work in Nodes 2.0.

A preset, palette, or tag preference is missing

  1. Check that the relevant file exists in ComfyUI-GadgetNodes/: models_config.yaml for model presets, train_config.yaml for tag-editor preferences, or a .yaml file inside prompt/ for the palette.
  2. Check spelling, indentation, and punctuation in the YAML file. Indentation determines which settings belong together.
  3. For model presets, check that the selected model name matches the pattern you wrote. Put specific patterns above general ones.
  4. For a palette, check that the final values in its category tree are prompt strings.

Wildcard text is not expanding

  1. Check that the wildcard .txt file is in ComfyUI/wildcards/ or ComfyUI-GadgetNodes/wildcards/.
  2. Check that the reference in your prompt matches the filename: colors.txt is referenced as {__colors__}.
  3. If you recently changed the file, check the node’s auto_refresh setting.
  4. If both wildcard folders exist, remember that ComfyUI/wildcards/ has priority.

Frequently Asked Questions

Does ComfyUI-GadgetNodes include checkpoints, LoRAs, or other image-generation models?
No models are listed as part of the extension. Its model nodes load or organize settings and metadata for models you use separately.

Do I need to create configuration files before using any nodes?
No. The extension can run without them, but features that depend on saved presets, palette entries, or tag-editor preferences will be limited.

Why did my preset use a value from an unexpected pattern?
Your model name may match multiple patterns. The first matching pattern that supplies a setting takes priority; later matches fill missing settings only. Check their order in models_config.yaml.

Does Virtual Group change what my workflow generates?
No. It changes how specified nodes appear on the canvas, not how the workflow executes.

Where do metadata-editor notes go?
They are saved in a .json file beside the corresponding LoRA, checkpoint, or diffusion model file.

Learn More about ComfyUI-GadgetNodes

ComfyUI-GadgetNodes Related Nodes

RunComfy
Copyright 2025 RunComfy. All Rights Reserved.

RunComfy is the premier ComfyUI platform, offering ComfyUI online environment and services, along with ComfyUI workflows featuring stunning visuals. RunComfy also provides AI Models, enabling artists to harness the latest AI tools to create incredible art.

ComfyUI-GadgetNodes detailed guide | ComfyUI