Comfyui

🚀 ComfyUI MCP Server

A lightweight Python-based MCP (Model Context Protocol) server that interfaces with a local ComfyUI instance to generate images upon requests from AI agents.

🚀 Quick Start

This project enables AI agents to send image generation requests to ComfyUI via WebSocket using the MCP protocol. It supports flexible workflow selection, dynamic parameters, and returns image URLs provided by ComfyUI.

✨ Features

• Flexible Workflow Selection: Choose from various workflows (e.g., basic_api_test.json).
• Dynamic Parameters: Support parameters like prompt, width, height, and model.
• Image URL Return: Receive image URLs generated by ComfyUI.

📦 Installation

Prerequisites

• Python 3.10+
• ComfyUI: Installed and running locally (e.g., at localhost:8188).
• Dependencies: requests, websockets, mcp (install via pip).

Steps

1. Clone the Repository:

   git clone <your-repo-url>
   cd comfyui-mcp-server
   

2. Install Dependencies:

   pip install requests websockets mcp
   

3. Start ComfyUI:
   • Install ComfyUI (refer to ComfyUI documentation).
   • Run it on port 8188:

     cd <ComfyUI_dir>
     python main.py --port 8188
     

4. Prepare Workflows:
   • Place API - formatted workflow files (e.g., basic_api_test.json) in the workflows/ directory.
   • Export workflows using ComfyUI's UI with the "Save (API format)" option enabled in the development mode settings.

💻 Usage Examples

Basic Usage

1. Run the MCP Server:

   python server.py
   

   • It listens on ws://localhost:9000.
2. Use the Test Client:

   python client.py
   

   • Send a sample request: "a dog wearing sunglasses", with a size of 512x512 and the model sd_xl_base_1.0.safetensors.
   • Sample output:

     {
       "image_url": "http://localhost:8188/view?filename=ComfyUI_00001_.png&subfolder=&type=output"
     }
     

Advanced Usage

Customize Requests:

• Modify the payload in client.py to change prompt, width, height, workflow_id, or model.
• Example:

  "params": json.dumps({
      "prompt": "a cat in space",
      "width": 768,
      "height": 768,
      "workflow_id": "basic_api_test",
      "model": "v1-5-pruned-emaonly.ckpt"
  })
  

📚 Documentation

Project Structure

• server.py: The MCP server with WebSocket transport and lifecycle support.
• comfyui_client.py: Interfaces with ComfyUI's API and handles workflow queuing.
• client.py: A test client for sending MCP requests.
• workflows/: A directory for storing API - formatted workflow JSON files.

Notes

⚠️ Important Note

• Ensure that the selected model (e.g., v1-5-pruned-emaonly.ckpt) exists in <ComfyUI_dir>/models/checkpoints/.
• The MCP SDK does not have native WebSocket transport; this implementation uses a custom one.
• For custom workflows, you can adjust the DEFAULT_MAPPING node ID in comfyui_client.py if necessary.

Contribution

Welcome to submit issues or PRs to enhance flexibility (e.g., dynamic node mapping, progress streaming).

📄 License

This project is under the Apache License.