MCP Image Generation Server

An MCP (Model Context Protocol) server that provides AI image generation capabilities through multiple providers including OpenAI DALL-E and Replicate.

Features

• Multi-provider support: OpenAI (DALL-E 2/3) and Replicate (Stable Diffusion, etc.)
• Flexible configuration: Configure providers via environment variables
• Size handling: Automatic size conversion for different providers
• Model selection: Choose specific models per request
• Error handling: Graceful error handling with informative messages

Installation

From npm (Recommended)

npm install @renzof/mcp-image-gen

From Source

Alternatively, you can clone and build locally:

git clone https://github.com/RenzoF/mcp-image-gen.git
cd mcp-image-gen
npm install
npm run build

Configuration

API Keys

API keys can be provided in two ways:

1. Through MCP configuration (recommended for Claude Desktop):

   • Pass API keys as command-line arguments in the MCP configuration
   • Keys are securely passed when the server starts

2. Through environment variables (for development/testing):

   • Set OPENAI_API_KEY for OpenAI
   • Set REPLICATE_API_TOKEN for Replicate

Claude Desktop Configuration

Add to your Claude Desktop configuration file:

If installed via npm:

{
  "mcpServers": {
    "image-gen": {
      "command": "npx",
      "args": [
        "@renzof/mcp-image-gen",
        "--openai-api-key",
        "sk-your-openai-api-key",
        "--replicate-api-token",
        "r8_your-replicate-token"
      ]
    }
  }
}

If cloned locally:

{
  "mcpServers": {
    "image-gen": {
      "command": "node",
      "args": [
        "/path/to/mcp-image-gen/dist/index.js",
        "--openai-api-key",
        "sk-your-openai-api-key",
        "--replicate-api-token",
        "r8_your-replicate-token"
      ]
    }
  }
}

Notes:

• Replace the API key placeholders with your actual API keys
• At least one provider API key is required (OpenAI or Replicate)
• For local installation, replace /path/to/mcp-image-gen with your actual path
• The npm package will be automatically downloaded when first used with npx

Usage

The server provides a generate_image tool with the following parameters:

Parameters

• prompt (required): Text description of the image to generate
• size (optional): Image dimensions (default: "1024x1024")
  • Common sizes: "256x256", "512x512", "1024x1024", "1792x1024", "1024x1792"
• provider (optional): Image generation provider ("openai" or "replicate")
• model (optional): Specific model to use
  • OpenAI: "dall-e-2", "dall-e-3" (default)
  • Replicate: Model version string (e.g., "stability-ai/stable-diffusion:...")

Example Usage in Claude

Generate an image of a sunset over mountains

Generate an image with size 1792x1024 of a futuristic city

Generate an image using replicate provider of a cat wearing sunglasses

Supported Providers

OpenAI

• Models: DALL-E 2, DALL-E 3
• Sizes: 256x256, 512x512, 1024x1024, 1792x1024, 1024x1792
• Requires: OPENAI_API_KEY

Replicate

• Models: Stable Diffusion and other models on Replicate
• Sizes: Any custom size (provider handles dynamically)
• Requires: REPLICATE_API_TOKEN

Development

Building

npm install
npm run build

Running locally

npm run dev

Project Structure

mcp-image-gen/
├── src/
│   ├── index.ts          # MCP server implementation
│   ├── image-generator.ts # Image generation logic
│   └── types.ts          # TypeScript type definitions
├── dist/                 # Compiled JavaScript (generated)
├── package.json
├── tsconfig.json
└── README.md

Error Handling

The server handles various error cases:

• Missing API keys
• Invalid provider selection
• API errors from providers
• Network issues
• Invalid parameters

Errors are returned with descriptive messages to help troubleshoot issues.

License

MIT