# Comfy UI
**Action ID:** `comfy_ui`
## Description
Run any ComfyUI workflow.
## Connection
| Name | Description | Required | Category |
| ------------------ | ------------------------------------------- | -------- | -------- |
| PixelML Connection | The PixelML connection to call PixelML API. | True | pixelml |
## Input Parameters
| Name | Type | Required | Default | Description |
| -------------- | ------ | :------: | --------------------- | --------------------------------------------------------------------------------------- |
| workflow\_json | string | - | Default SDXL workflow | Your ComfyUI workflow in API format JSON. Export from ComfyUI using 'Save (API format)' |
View JSON Schema
```json
{
"description": "Comfy UI node input.",
"properties": {
"workflow_json": {
"default": "{\"3\":{\"inputs\":{\"seed\":156680208700286,\"steps\":10,\"cfg\":2.5,\"sampler_name\":\"dpmpp_2m_sde\",\"scheduler\":\"karras\",\"denoise\":1,\"model\":[\"4\",0],\"positive\":[\"6\",0],\"negative\":[\"7\",0],\"latent_image\":[\"5\",0]},\"class_type\":\"KSampler\",\"_meta\":{\"title\":\"KSampler\"}},\"4\":{\"inputs\":{\"ckpt_name\":\"SDXL-Flash.safetensors\"},\"class_type\":\"CheckpointLoaderSimple\",\"_meta\":{\"title\":\"Load Checkpoint\"}},\"5\":{\"inputs\":{\"width\":1024,\"height\":1024,\"batch_size\":1},\"class_type\":\"EmptyLatentImage\",\"_meta\":{\"title\":\"Empty Latent Image\"}},\"6\":{\"inputs\":{\"text\":[\"10\",0],\"clip\":[\"4\",1]},\"class_type\":\"CLIPTextEncode\",\"_meta\":{\"title\":\"CLIP Text Encode (Prompt)\"}},\"7\":{\"inputs\":{\"text\":[\"11\",0],\"clip\":[\"4\",1]},\"class_type\":\"CLIPTextEncode\",\"_meta\":{\"title\":\"CLIP Text Encode (Prompt)\"}},\"8\":{\"inputs\":{\"samples\":[\"3\",0],\"vae\":[\"4\",2]},\"class_type\":\"VAEDecode\",\"_meta\":{\"title\":\"VAE Decode\"}},\"9\":{\"inputs\":{\"filename_prefix\":\"ComfyUI\",\"images\":[\"8\",0]},\"class_type\":\"SaveImage\",\"_meta\":{\"title\":\"Save Image\"}},\"10\":{\"inputs\":{\"name\":\"prompt\",\"value\":\"beautiful scenery nature glass bottle landscape, purple galaxy bottle,\"},\"class_type\":\"StringInput_PixelML\",\"_meta\":{\"title\":\"StringInput_PixelML\"}},\"11\":{\"inputs\":{\"name\":\"negative_prompt\",\"value\":\"text, watermark\"},\"class_type\":\"StringInput_PixelML\",\"_meta\":{\"title\":\"StringInput_PixelML\"}}}",
"description": "Your ComfyUI workflow as JSON. You must use the API version of your workflow. Get it from ComfyUI using 'Save (API format)'",
"title": "Workflow JSON (API Format)",
"type": "string"
}
},
"title": "ComfyUINodeInput",
"type": "object"
}
```
## Output Parameters
| Name | Type | Description |
| ------- | ----- | ----------------------------------------------------- |
| outputs | array | Array of image URLs generated by the ComfyUI workflow |
View JSON Schema
```json
{
"description": "Comfy UI node output.",
"properties": {
"outputs": {
"items": {
"title": "Output",
"type": "string"
},
"title": "List outputs",
"type": "array"
}
},
"required": [
"outputs"
],
"title": "ComfyUINodeOutput",
"type": "object"
}
```
## How It Works
This node executes custom ComfyUI workflows through the PixelML API, allowing you to leverage ComfyUI's powerful node-based image generation capabilities within your AgenticFlow workflows. You provide your ComfyUI workflow in API JSON format (exported from ComfyUI), and the node processes it remotely, returning URLs to the generated images for further use in your workflow pipeline.
## Usage Examples
### Example 1: Basic Text-to-Image Generation
**Input:**
```
workflow_json: {
"3": {
"inputs": {
"seed": 42,
"steps": 20,
"cfg": 7.0,
"sampler_name": "euler",
"scheduler": "normal",
"denoise": 1,
"model": ["4", 0],
"positive": ["6", 0],
"negative": ["7", 0],
"latent_image": ["5", 0]
},
"class_type": "KSampler"
},
"10": {
"inputs": {
"name": "prompt",
"value": "professional portrait of a business woman in modern office, natural lighting"
},
"class_type": "StringInput_PixelML"
}
// ... additional nodes
}
```
**Output:**
```
outputs: [
"https://storage.pixelml.com/outputs/comfyui_12345_0.png"
]
```
### Example 2: Batch Image Generation
**Input:**
```
workflow_json: {
"5": {
"inputs": {
"width": 512,
"height": 768,
"batch_size": 4
},
"class_type": "EmptyLatentImage"
},
"10": {
"inputs": {
"name": "prompt",
"value": "fantasy landscape with mountains and rivers, concept art style"
},
"class_type": "StringInput_PixelML"
}
// ... additional nodes
}
```
**Output:**
```
outputs: [
"https://storage.pixelml.com/outputs/comfyui_67890_0.png",
"https://storage.pixelml.com/outputs/comfyui_67890_1.png",
"https://storage.pixelml.com/outputs/comfyui_67890_2.png",
"https://storage.pixelml.com/outputs/comfyui_67890_3.png"
]
```
### Example 3: Image-to-Image with ControlNet
**Input:**
```
workflow_json: {
"12": {
"inputs": {
"image": "https://example.com/input-sketch.png",
"control_net": "control_v11p_sd15_canny"
},
"class_type": "ControlNetLoader"
},
"10": {
"inputs": {
"name": "prompt",
"value": "realistic photo based on sketch, professional photography"
},
"class_type": "StringInput_PixelML"
}
// ... additional nodes
}
```
**Output:**
```
outputs: [
"https://storage.pixelml.com/outputs/comfyui_24680_0.png"
]
```
## Common Use Cases
* **Custom AI Image Generation**: Execute complex ComfyUI workflows with custom models, LoRAs, and processing chains
* **Automated Content Creation**: Generate product images, marketing visuals, or social media content at scale
* **Image Processing Pipelines**: Apply sophisticated image transformations like upscaling, inpainting, or style transfer
* **Batch Image Production**: Generate multiple variations of images with different seeds or parameters in one workflow run
* **ControlNet Integration**: Use advanced conditioning methods like pose, depth, or canny edge detection for precise image generation
* **Custom Model Workflows**: Leverage specialized fine-tuned models or custom trained checkpoints not available in standard tools
* **Complex Multi-Stage Generation**: Chain multiple generation steps, refinements, and post-processing in a single workflow
## Error Handling
| Error Type | Cause | Solution |
| -------------------------- | ------------------------------------------------------ | ------------------------------------------------------------------------ |
| Invalid JSON Format | workflow\_json is not valid JSON | Ensure the workflow is properly formatted JSON from ComfyUI's API export |
| Missing Required Nodes | Workflow missing essential nodes like SaveImage | Include all necessary nodes; ensure at least one SaveImage node exists |
| Invalid Node Configuration | Node inputs reference non-existent nodes or outputs | Verify all node connections and references are correct in the workflow |
| Model Not Found | Referenced checkpoint or model doesn't exist on server | Use models available in the PixelML environment or check model names |
| Insufficient Resources | Workflow requires more GPU memory than available | Reduce batch size, image resolution, or simplify the workflow |
| Connection Timeout | PixelML API connection failed or timed out | Check your PixelML connection settings and network connectivity |
| Execution Failed | ComfyUI workflow execution encountered an error | Review the workflow for errors; test locally in ComfyUI first |
## Notes
* **PixelML Connection Required**: This node requires a valid PixelML connection to execute ComfyUI workflows remotely.
* **API Format Only**: You must export your workflow using ComfyUI's 'Save (API format)' option, not the regular workflow format.
* **StringInput\_PixelML Nodes**: Use StringInput\_PixelML nodes in your workflow to make it dynamic and accept parameters from AgenticFlow.
* **Model Availability**: Only models and custom nodes available in the PixelML environment can be used; not all local custom nodes may be supported.
* **Output Image Storage**: Generated images are temporarily stored and accessible via URLs; download or save them for long-term use.
* **Workflow Complexity**: Complex workflows with many nodes may take longer to execute; consider execution time in your workflow design.
* **Testing**: Always test your ComfyUI workflow locally before integrating it into AgenticFlow to ensure it works correctly.
* **Batch Processing**: The outputs array will contain multiple URLs if your workflow generates multiple images via batch\_size or multiple SaveImage nodes.
---
# Agent Instructions: Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.
Perform an HTTP GET request on the current page URL with the `ask` query parameter:
```
GET https://docs.agenticflow.ai/reference/nodes/comfy_ui.md?ask=
```
The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.
Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.