comfy-mcp-pipeline
by: lalanikarim
This is a pipeline wrapper for comfy-mcp-server for Open WebUI.
📌Overview
Purpose: The Comfy MCP Pipeline serves as a wrapper for the Comfy MCP Server, facilitating seamless integration with Open WebUI pipelines for image generation workflows.
Overview: The Comfy MCP Pipeline leverages the functionalities of the Comfy MCP Server to enable users to execute image generation tasks through an intuitive chat interface. By connecting to predefined workflows, the pipeline streamlines the process of utilizing Comfy UI features and enhances accessibility for users aiming to generate images based on specified prompts.
Key Features:
-
Integration with Comfy MCP Server: Provides a direct interface to operate the Comfy MCP Server within Open WebUI, simplifying workflow management and execution.
-
Customizable Workflow Handling: Supports the import and setup of custom workflows from Comfy UI, allowing users to tailor image generation processes based on specific project requirements.
Comfy MCP Pipeline
This is a wrapper for the Comfy MCP Server to be used with Open WebUI.
Prerequisites
- Open WebUI
- Open WebUI Pipelines
- ComfyUI
- Updated
requirements.txt
for the pipelines server - JSON Export of a ComfyUI Workflow API
Pipeline Installation and Setup
- Upload the
comfy-mcp-pipeline.py
to the Pipeline server according to the Open WebUI Pipelines documentation. - Choose
comfy-mcp-pipeline (pipe)
fromPipeline Valves
. - Set the configuration for the following valves:
- Comfy Url: URL for your Comfy UI server.
- Comfy Url External: External URL for your Comfy UI server. Use
Comfy Url
value if same. - Comfy Workflow Json File: Path of the workflow JSON file.
- Prompt Node Id: ID of the text prompt node from the workflow JSON file.
- Output Node Id: ID of the generated image node from the workflow JSON file.
- If all steps are successful, you will see
Comfy MCP Pipeline
in the list of models.
Usage
- Select
New Chat
and selectComfy MCP Pipeline
. - Enter an image generation prompt and hit send.
- If the setup was successful, you should see the generated image.