README.md•3.5 kB
# ComfyUI API Integration
This module provides integration with ComfyUI's API for the Createve.AI API Server, allowing execution of ComfyUI workflows and retrieval of workflow information.
## Features
- Execute ComfyUI workflows with customizable parameters
- Get information about available workflows
- Support for both text-to-image and image-to-image workflows
- Configuration through YAML file
- Async execution with progress monitoring
- MCP integration for AI assistant access
## Installation
1. Ensure ComfyUI is running and accessible (default: http://localhost:8188)
2. Install required dependencies:
```bash
pip install -r requirements.txt
```
## Configuration
The module uses a `config.yaml` file for configuration:
```yaml
url: http://localhost:8188 # ComfyUI server URL
workflows_directory: workflows # Directory for workflow files
default_timeout: 300 # Default execution timeout
workflows:
txt2img: # Example workflow configuration
file: txt2img.json
timeout: 600
default_params:
prompt: "A beautiful landscape"
negative_prompt: "blur, deformed"
width: 512
height: 512
steps: 20
cfg_scale: 7.0
seed: -1
```
## Usage
### As REST API
Execute a workflow:
```bash
curl -X POST "http://localhost:43080/api/comfyui/comfyUIWorkflowExecutor" \
-H "Content-Type: application/json" \
-d '{
"workflow_name": "txt2img",
"prompt": "A beautiful sunset over mountains",
"negative_prompt": "blur, noise",
"width": 768,
"height": 512
}'
```
Get workflow information:
```bash
curl -X POST "http://localhost:43080/api/comfyui/comfyUIWorkflowInfo" \
-H "Content-Type: application/json" \
-d '{
"workflow_name": "txt2img"
}'
```
### As MCP Resource
AI assistants can interact with the ComfyUI integration through MCP:
```python
# Example MCP tool usage
response = await mcp.call_tool(
"comfyui_workflowExecutor",
{
"workflow_name": "txt2img",
"prompt": "A beautiful sunset over mountains",
"width": 768,
"height": 512
}
)
```
## Workflow Files
Workflow files (`.json`) should be placed in the configured workflows directory. Each workflow file should be a valid ComfyUI workflow JSON export.
Example workflow structure:
```json
{
"nodes": {
"1": {
"class_type": "CLIPTextEncode",
"inputs": {
"prompt": "A beautiful landscape",
"negative_prompt": "blur, deformed"
}
},
"2": {
"class_type": "KSampler",
"inputs": {
"width": 512,
"height": 512,
"steps": 20,
"cfg_scale": 7.0,
"seed": -1
}
}
}
}
```
## Error Handling
The module includes comprehensive error handling:
- Configuration errors (missing files, invalid YAML)
- Workflow execution errors (timeout, API errors)
- Parameter validation
- ComfyUI server connection issues
Error responses include detailed messages to help diagnose issues.
## Limitations
- Currently supports basic parameter mapping for common node types
- Limited to single-image output workflows
- Does not support all ComfyUI node types
- Queue mode required for execution (may have latency)
## Development
To add new features or modify existing ones:
1. Update workflow parameter mapping in `api.py`
2. Add new configuration options in `config.py`
3. Update input/output types in API classes
4. Test with different workflow configurations
## License
MIT License - See LICENSE file for details