🏃‍♂️ Quick Start

Choose the deployment method that best suits your needs, from simple to complex:

🎯 Method 1: One-click Experience

💡 Zero configuration startup, perfect for quick experience and testing

🚀 Temporary Run

# First you need to install the uv environment
# Start with one command, no system installation required
uvx pixelle@latest

📚 View uvx CLI Reference →

📦 Persistent Installation

# Here you need to install it in the python3.11 environment
# Install to system
pip install -U pixelle

# Start service
pixelle

📚 View pip CLI Reference →

After startup, it will automatically enter the configuration wizard to guide you through execution engine selection (ComfyUI/RunningHub) and LLM configuration.

🛠️ Method 2: Local Development Deployment

💡 Supports custom workflows and secondary development

📥 1. Get Source Code

git clone https://github.com/AIDC-AI/Pixelle-MCP.git
cd Pixelle-MCP

🚀 2. Start Service

# Interactive mode (recommended)
uv run pixelle

📚 View Complete CLI Reference →

🔧 3. Add Custom Workflows (Optional)

# Copy example workflows to data directory (run this in your desired project directory)
cp -r workflows/* ./data/custom_workflows/

⚠️ Important: Make sure to test workflows in ComfyUI first to ensure they run properly, otherwise execution will fail.

🐳 Method 3: Docker Deployment

💡 Suitable for production environments and containerized deployment

📋 1. Prepare Configuration

git clone https://github.com/AIDC-AI/Pixelle-MCP.git
cd Pixelle-MCP

# Create environment configuration file
cp .env.example .env
# Edit .env file to configure your ComfyUI address and LLM settings

🚀 2. Start Container

# Start all services in background
docker compose up -d

# View logs
docker compose logs -f

🌐 Access Services

Regardless of which method you use, after startup you can access via:

• 🌐 Web Interface: http://localhost:9004
  Default username and password are both dev, can be modified after startup
• 🔌 MCP Endpoint: http://localhost:9004/pixelle/mcp
  For MCP clients like Cursor, Claude Desktop to connect

💡 Port Configuration: Default port is 9004, can be customized via environment variable PORT=your_port.

⚙️ Initial Configuration

On first startup, the system will automatically detect configuration status:

1. 🚀 Execution Engine Selection: Choose between local ComfyUI or RunningHub cloud service
2. 🤖 LLM Configuration: Configure at least one LLM provider (OpenAI, Ollama, etc.)
3. 📁 Workflow Directory: System will automatically create necessary directory structure

🌐 RunningHub Cloud Mode Advantages

• ✅ Zero Hardware Requirements: No need for local GPU or high-performance hardware
• ✅ No Environment Setup: No need to install and configure ComfyUI locally
• ✅ Ready to Use: Register and get API key to start immediately
• ✅ Stable Performance: Professional cloud infrastructure ensures stable execution
• ✅ Auto Scaling: Automatically handles concurrent requests and resource allocation

🏠 Local ComfyUI Mode Advantages

• ✅ Full Control: Complete control over execution environment and model versions
• ✅ Privacy Protection: All data processing happens locally, ensuring data privacy
• ✅ Custom Models: Support for custom models and nodes not available in cloud
• ✅ No Network Dependency: Can work offline without internet connection
• ✅ Cost Control: No cloud service fees for high-frequency usage

🆘 Need Help? Join community groups for support (see Community section below)

🛠️ Add Your Own MCP Tool

⚡ One workflow = One MCP Tool, supports two addition methods:

📋 Method 1: Local ComfyUI Workflow - Export API format workflow files 📋 Method 2: RunningHub Workflow ID - Use cloud workflow IDs directly

🎯 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)

Note: When using RunningHub, you only need to input the corresponding workflow ID, no need to download and upload workflow files.

🔧 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:

$<param_name>.[~]<field_name>[!][:<description>]

🔍 Syntax Explanation:

• param_name: The parameter name for the generated MCP tool function
• ~: Optional, indicates URL parameter upload processing, returns relative path
• field_name: The corresponding input field in the node
• !: Indicates this parameter is required
• description: 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's image 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

🎯 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