ComfyUI Extension: ComfyUI Prompt Saver
Run ComfyUI workflows without the setup
No installs, no CUDA version roulette, no GPU sitting idle on your bill. Bring a workflow and run it in the browser.
Saves images with prompt reference cards - thumbnail + formatted prompt side-by-side with JSON syntax highlighting for easy browsing and organization.
README
ComfyUI Prompt Saver
A ComfyUI custom node that saves images alongside formatted prompt reference cards for easy browsing and organization.
Features
- Reference Cards: Creates side-by-side images with thumbnail + formatted prompt text
- JSON Syntax Highlighting: Automatically highlights JSON prompts with color-coded keys, values, and braces
- Multiple Output Formats: Save full image, reference card, and/or JSON metadata
- Auto Text Sizing: Dynamically adjusts font size to fit prompts in available space
- Prompt Cleaning: Handles escaped characters and JSON parsing automatically
Example Output
The reference card displays your generated image as a thumbnail alongside the formatted prompt:
┌─────────────────────────────────────────────────┐
│ ┌──────────┐ { │
│ │ │ "scene": "mountain landscape", │
│ │ Image │ "style": "oil painting", │
│ │ Thumbnail│ "lighting": "golden hour" │
│ │ │ } │
│ └──────────┘ │
└─────────────────────────────────────────────────┘
Installation
Via ComfyUI Manager (Recommended)
- Open ComfyUI Manager
- Search for "Prompt Saver"
- Click Install
Manual Installation
cd ComfyUI/custom_nodes
git clone https://github.com/robomello/comfyui-prompt-saver
Restart ComfyUI after installation.
Usage
- Find the node under
image/savecategory as "Save Image + Prompt Reference" - Connect your image output and prompt string
- Configure options as needed
Inputs
| Input | Type | Description |
|-------|------|-------------|
| images | IMAGE | Required. The generated image(s) to save |
| prompt | STRING | Required. The prompt text (supports plain text or JSON) |
| filename_prefix | STRING | Prefix for saved files (default: "prompt_ref") |
| thumbnail_size | INT | Size of thumbnail in reference card (128-1024, default: 512) |
| save_full_image | BOOLEAN | Save the full-resolution image (default: True) |
| save_reference_card | BOOLEAN | Save the reference card with thumbnail + prompt (default: True) |
| save_json | BOOLEAN | Save prompt metadata as JSON file (default: True) |
| negative_prompt | STRING | Optional. Negative prompt to include in metadata |
| extra_metadata | STRING | Optional. Additional metadata to include |
Outputs
| Output | Type | Description |
|--------|------|-------------|
| images | IMAGE | Passthrough of input images |
| saved_path | STRING | Comma-separated list of saved file paths |
Output Files
Files are saved to ComfyUI/output/prompt_references/ with the following naming:
{prefix}_{timestamp}_{index}_full.png- Full resolution image{prefix}_{timestamp}_{index}_ref.png- Reference card{prefix}_{timestamp}_{index}.json- Prompt metadata
Requirements
- ComfyUI
- Pillow
- numpy
License
MIT License - see LICENSE
Author
Roberto de Mello (@robomello)
Run ComfyUI workflows without the setup
No installs, no CUDA version roulette, no GPU sitting idle on your bill. Bring a workflow and run it in the browser.