Community platform integration mentioned for user support and updates related to the MCP server
Supports containerized deployment of the MCP server services for easier setup and management
Integrates with Flux models for advanced image generation capabilities through ComfyUI workflows
Integrates with Ollama for local LLM support, allowing AI agents to use locally hosted models for controlling ComfyUI workflows and AIGC operations
Provides support for OpenAI models through the LiteLLM framework, enabling AI agents to leverage OpenAI's capabilities for multimodal content generation using ComfyUI workflows
Community platform integration mentioned for user support and updates related to the MCP server
https://github.com/user-attachments/assets/65422cef-96f9-44fe-a82b-6a124674c417
📋 Recent Updates
- ✅ 2025-08-12: Integrated the LiteLLM framework, adding multi-model support for Gemini, DeepSeek, Claude, Qwen, and more
🚀 Features
- ✅ 🔄 Full-modal Support: Supports TISV (Text, Image, Sound/Speech, Video) full-modal conversion and generation
- ✅ 🧩 ComfyUI Ecosystem: Server-side is built on ComfyUI, inheriting all capabilities from the open ComfyUI ecosystem
- ✅ 🔧 Zero-code Development: Defines and implements the Workflow-as-MCP Tool solution, enabling zero-code development and dynamic addition of new MCP Tools
- ✅ 🗄️ MCP Server: Server provides functionality based on the MCP protocol, supporting integration with any MCP client (including but not limited to Cursor, Claude Desktop, etc.)
- ✅ 🌐 MCP Client: Client is developed based on the Chainlit framework, inheriting Chainlit's UI controls and supporting integration with more MCP Servers
- ✅ 🔄 Flexible Deployment: Supports standalone deployment of Server-side only as MCP Server, or standalone deployment of Client-side only as MCP Client, or combined deployment
- ✅ ⚙️ Unified Configuration: Uses YAML configuration scheme, one config file manages all services
- ✅ 🤖 Multi-LLM Support: Supports multiple mainstream LLMs, including OpenAI, Ollama, Gemini, DeepSeek, Claude, Qwen, and more
📁 Project Structure
- mcp-base: 🔧 Basic service, provides file storage and shared service capabilities
- mcp-client: 🌐 MCP client, a web interface built on Chainlit
- mcp-server: 🗄️ MCP server, provides various AIGC tools and services
🏃♂️ Quick Start
📥 1. Clone the Source Code & Configure Services
📦 1.1 Clone the Source Code
⚙️ 1.2 Configure Services
The project uses a unified YAML configuration scheme:
📋 Detailed Configuration Instructions:
The configuration file contains three main sections: Basic Service, MCP Server, and MCP Client. Each section has detailed configuration item descriptions in config.yml.example
.
🔍 Configuration Checklist:
- ✅ Copied
config.yml.example
toconfig.yml
- ✅ Configured ComfyUI service address (ensure ComfyUI is running)
- ✅ Configured at least one LLM model (OpenAI or Ollama)
- ✅ Port numbers are not occupied by other services (9001, 9002, 9003)
🔧 2. Add MCP Tool (Optional)
This step is optional and only affects your Agent's capabilities. You can skip it if not needed for now.
The mcp-server/workflows
directory contains a set of popular workflows by default. Run the following command to copy them to your mcp-server. When the service starts, they will be automatically converted into MCP Tools for LLM use.
Note: It is strongly recommended to test the workflow in your ComfyUI canvas before copying, to ensure smooth execution later.
🚀 3. Start the Services
🎯 3.1 Start with Docker (Recommended)
🛠️ 3.2 One-click Script Start
Requires uv environment.
Linux/macOS users:
Windows users:
Simply double-click the run.bat
script in the root directory
🛠️ 3.3 Manual Service Start
Requires uv environment.
Start Basic Service (mcp-base):
Start Server (mcp-server):
Start Client (mcp-client):
🌐 4. Access the Services
After startup, the service addresses are as follows:
- Client: 🌐 http://localhost:9003 (Chainlit Web UI, default username and password are both
dev
, can be changed inauth.py
) - Server: 🗄️ http://localhost:9002/sse (MCP Server)
- Base Service: 🔧 http://localhost:9001/docs (File storage and basic API)
🛠️ Add Your Own MCP Tool
⚡ One workflow = One MCP Tool
🎯 1. Add the Simplest MCP Tool
- 📝 Build a workflow in ComfyUI for image Gaussian blur (Get it here), then set the
LoadImage
node's title to$image.image!
as shown below: - 📤 Export it as an API format file and rename it to
i_blur.json
. You can export it yourself or use our pre-exported version (Get it here) - 📋 Copy the exported API workflow file (must be API format), input it on the web page, and let the LLM add this Tool
- ✨ After sending, the LLM will automatically convert this workflow into an MCP Tool
- 🎨 Now, refresh the page and send any image to perform Gaussian blur processing via LLM
🔌 2. Add a Complex MCP Tool
The steps are the same as above, only the workflow part differs (Download workflow: UI format and API format)
🔧 ComfyUI Workflow Custom Specification
🎨 Workflow Format
The system supports ComfyUI workflows. Just design your workflow in the canvas and export it as API format. Use special syntax in node titles to define parameters and outputs.
📝 Parameter Definition Specification
In the ComfyUI canvas, double-click the node title to edit, and use the following DSL syntax to define parameters:
🔍 Syntax Explanation:
param_name
: The parameter name for the generated MCP tool function~
: Optional, indicates URL parameter upload processing, returns relative pathfield_name
: The corresponding input field in the node!
: Indicates this parameter is requireddescription
: Description of the parameter
💡 Example:
Required parameter example:
- Set LoadImage node title to:
$image.image!:Input image URL
- Meaning: Creates a required parameter named
image
, mapped to the node'simage
field
URL upload processing example:
- Set any node title to:
$image.~image!:Input image URL
- Meaning: Creates a required parameter named
image
, system will automatically download URL and upload to ComfyUI, returns relative path
📝 Note:
LoadImage
,VHS_LoadAudioUpload
,VHS_LoadVideo
and other nodes have built-in functionality, no need to add~
marker
Optional parameter example:
- Set EmptyLatentImage node title to:
$width.width:Image width, default 512
- Meaning: Creates an optional parameter named
width
, mapped to the node'swidth
field, default value is 512
🎯 Type Inference Rules
The system automatically infers parameter types based on the current value of the node field:
- 🔢
int
: Integer values (e.g. 512, 1024) - 📊
float
: Floating-point values (e.g. 1.5, 3.14) - ✅
bool
: Boolean values (e.g. true, false) - 📝
str
: String values (default type)
📤 Output Definition Specification
🤖 Method 1: Auto-detect Output Nodes
The system will automatically detect the following common output nodes:
- 🖼️
SaveImage
- Image save node - 🎬
SaveVideo
- Video save node - 🔊
SaveAudio
- Audio save node - 📹
VHS_SaveVideo
- VHS video save node - 🎵
VHS_SaveAudio
- VHS audio save node
🎯 Method 2: Manual Output Marking
Usually used for multiple outputs Use
$output.var_name
in any node title to mark output:
- Set node title to:
$output.result
- The system will use this node's output as the tool's return value
📄 Tool Description Configuration (Optional)
You can add a node titled MCP
in the workflow to provide a tool description:
- Add a
String (Multiline)
or similar text node (must have a single string property, and the node field should be one of: value, text, string) - Set the node title to:
MCP
- Enter a detailed tool description in the value field
⚠️ Important Notes
- 🔒 Parameter Validation: Optional parameters (without !) must have default values set in the node
- 🔗 Node Connections: Fields already connected to other nodes will not be parsed as parameters
- 🏷️ Tool Naming: Exported file name will be used as the tool name, use meaningful English names
- 📋 Detailed Descriptions: Provide detailed parameter descriptions for better user experience
- 🎯 Export Format: Must export as API format, do not export as UI format
💬 Community
Scan the QR codes below to join our communities for latest updates and technical support:
Discord Community | WeChat Group |
---|---|
🤝 How to Contribute
We welcome all forms of contribution! Whether you're a developer, designer, or user, you can participate in the project in the following ways:
🐛 Report Issues
- 📋 Submit bug reports on the Issues page
- 🔍 Please search for similar issues before submitting
- 📝 Describe the reproduction steps and environment in detail
💡 Feature Suggestions
- 🚀 Submit feature requests in Issues
- 💭 Describe the feature you want and its use case
- 🎯 Explain how it improves user experience
🔧 Code Contributions
📋 Contribution Process
- 🍴 Fork this repo to your GitHub account
- 🌿 Create a feature branch:
git checkout -b feature/your-feature-name
- 💻 Develop and add corresponding tests
- 📝 Commit changes:
git commit -m "feat: add your feature"
- 📤 Push to your repo:
git push origin feature/your-feature-name
- 🔄 Create a Pull Request to the main repo
🎨 Code Style
- 🐍 Python code follows PEP 8 style guide
- 📖 Add appropriate documentation and comments for new features
🧩 Contribute Workflows
- 📦 Share your ComfyUI workflows with the community
- 🛠️ Submit tested workflow files
- 📚 Add usage instructions and examples for workflows
🙏 Acknowledgements
❤️ Sincere thanks to the following organizations, projects, and teams for supporting the development and implementation of this project.
License
This project is released under the MIT License (LICENSE, SPDX-License-identifier: MIT).
This server cannot be installed
hybrid server
The server is able to function both locally and remotely, depending on the configuration or use case.
The AIGC solution based on the MCP protocol seamlessly transforms the ComfyUI workflow into an MCP Tool with zero code, enabling LLMS and ComfyUI to join forces and focus on image generation tasks. qwen-image, flux, and FLUx-KREA can be used for image generation.
Related MCP Servers
- AsecurityAlicenseAqualityEnables the generation of images using Together AI's models through an MCP server, supporting customizable parameters such as model selection, image dimensions, and output directory.Last updated -7MIT License
- -securityAlicense-qualityThe Comfy MCP Server uses the FastMCP framework to generate images from prompts by interacting with a remote Comfy server, allowing automated image creation based on workflow configurations.Last updated -30MIT License
- AsecurityAlicenseAqualityRun AI workflows hosted on Glif.app via MCP, including ComfyUI-based image generators, meme generators, selfies, chained LLM calls, and moreLast updated -025MIT License
- AsecurityAlicenseAqualityA server that integrates ComfyUI with MCP, allowing users to generate images and download them through natural language interactions.Last updated -48Apache 2.0