OpenAI Image Generation MCP Server

A TypeScript-based Model Context Protocol (MCP) server for generating images using OpenAI's image generation models (gpt-image-1, gpt-image-1-mini, DALL-E 3, and DALL-E 2).

Quick Start with npx

The easiest way to use this MCP server is with npx (no installation required):

npx @lpenguin/openai-image-mcp

This will start the MCP server with stdio transport. You'll need to set the OPENAI_API_KEY environment variable:

OPENAI_API_KEY=your_api_key_here npx @lpenguin/openai-image-mcp

Installation

You can also install the package globally:

npm install -g @lpenguin/openai-image-mcp

Then run it with:

openai-image-mcp

Development Setup

For local development:

1. Clone this repository
2. Install dependencies:

   npm install
   

3. Create a .env file based on .env.example and add your OpenAI API key:

   OPENAI_API_KEY=your_openai_api_key_here
   

4. Build the TypeScript code:

   npm run build
   

Running the MCP Server

Start the MCP server:

npm start

Development

For development with automatic compilation:

npm run dev

MCP Server Configuration

To use this as an MCP server with Claude or other MCP clients, you need to add it to your MCP settings configuration file.

Using npx (Recommended)

The easiest way is to use npx directly in your MCP configuration:

{
  "mcpServers": {
    "image-generation": {
      "command": "npx",
      "args": ["-y", "@lpenguin/openai-image-mcp"],
      "env": {
        "OPENAI_API_KEY": "your_openai_api_key_here"
      },
      "disabled": false,
      "autoApprove": []
    }
  }
}

Using local installation

If you have the package installed globally or locally, you can use the direct command:

{
  "mcpServers": {
    "image-generation": {
      "command": "openai-image-mcp",
      "env": {
        "OPENAI_API_KEY": "your_openai_api_key_here"
      },
      "disabled": false,
      "autoApprove": []
    }
  }
}

Using from source (Development)

For local development with the cloned repository:

{
  "mcpServers": {
    "image-generation": {
      "command": "node",
      "args": ["dist/mcp-server.js"],
      "env": {
        "OPENAI_API_KEY": "your_openai_api_key_here"
      },
      "disabled": false,
      "autoApprove": []
    }
  }
}

Important: Replace "your_openai_api_key_here" with your actual OpenAI API key. The MCP server will use this key to authenticate with the OpenAI API.

Copy this configuration to your MCP settings file, typically located at:

VS Code

VS Code has native MCP support. Add the configuration to your workspace or user settings:

Location: .vscode/mcp.json (in your workspace root)

Using npx:

{
  "mcpServers": {
    "image-generation": {
      "command": "npx",
      "args": ["-y", "@lpenguin/openai-image-mcp"],
      "env": {
        "OPENAI_API_KEY": "your_openai_api_key_here"
      },
      "disabled": false,
      "autoApprove": []
    }
  }
}

Using local installation:

{
  "mcpServers": {
    "image-generation": {
      "command": "node",
      "args": ["/absolute/path/to/openai-image-mcp/dist/mcp-server.js"],
      "env": {
        "OPENAI_API_KEY": "your_openai_api_key_here"
      },
      "disabled": false,
      "autoApprove": []
    }
  }
}

Important:

• Replace /absolute/path/to/openai-image-mcp with the actual absolute path to your cloned repository
• Replace "your_openai_api_key_here" with your actual OpenAI API key

Claude Desktop

Location: ~/Library/Application Support/Claude/claude_desktop_config.json (macOS)

Use the npx configuration format shown above.

API Key Configuration

The MCP server requires an OpenAI API key to function. There are two ways to provide this key:

1. Environment Variable: If you're running the server directly with npm start, it will use the OPENAI_API_KEY from your .env file.

2. MCP Settings: When configuring the server in your MCP settings file, you need to provide the API key in the env section as shown above. This key will be passed to the server as an environment variable.

Usage

Once the MCP server is configured in Claude, you can use one of the four image generation tools to create images based on text prompts. Each tool is optimized for a specific OpenAI model.

Available Tools

1. generate_image_gpt - GPT-Image-1 Model

Generate images using OpenAI's gpt-image-1 model with advanced features like transparency and custom output formats.

Parameters:

• prompt (required): A text description of the desired image (max 32,000 characters)
• output (required): File path where the generated image should be saved (e.g., /path/to/image.png)
• size (optional): Image size. Options: "1024x1024", "1536x1024", "1024x1536", "auto". Default: "auto"
• quality (optional): Image quality. Options: "low", "medium", "high", "auto". Default: "auto"
• background (optional): Background transparency. Options: "transparent", "opaque", "auto". Default: "auto"
• output_format (optional): Output image format. Options: "png", "jpeg", "webp". Default: "png"
• output_compression (optional): Compression level (0-100) for webp/jpeg formats. Default: 100
• moderation (optional): Content moderation level. Options: "low", "auto". Default: "auto"
• n (optional): Number of images to generate (1-10). Default: 1

2. generate_image_gpt_mini - GPT-Image-1-Mini Model

Generate images using OpenAI's gpt-image-1-mini model. Cost-efficient alternative to gpt-image-1 with the same advanced features like transparency and custom output formats.

Parameters:

• prompt (required): A text description of the desired image (max 32,000 characters)
• output (required): File path where the generated image should be saved (e.g., /path/to/image.png)
• size (optional): Image size. Options: "1024x1024", "1536x1024", "1024x1536", "auto". Default: "auto"
• quality (optional): Image quality. Options: "low", "medium", "high", "auto". Default: "auto"
• background (optional): Background transparency. Options: "transparent", "opaque", "auto". Default: "auto"
• output_format (optional): Output image format. Options: "png", "jpeg", "webp". Default: "png"
• output_compression (optional): Compression level (0-100) for webp/jpeg formats. Default: 100
• moderation (optional): Content moderation level. Options: "low", "auto". Default: "auto"
• n (optional): Number of images to generate (1-10). Default: 1

3. generate_image_dalle3 - DALL-E 3 Model

Generate high-quality images using OpenAI's DALL-E 3 model with style control.

Parameters:

• prompt (required): A text description of the desired image (max 4,000 characters)
• output (required): File path where the generated image should be saved
• size (optional): Image size. Options: "1024x1024", "1792x1024", "1024x1792". Default: "1024x1024"
• quality (optional): Image quality. Options: "standard", "hd". Default: "standard"
• style (optional): Image style. Options: "vivid" (hyper-real and dramatic), "natural" (more natural, less hyper-real). Default: "vivid"

Note: DALL-E 3 can only generate 1 image at a time (n is always 1).

4. generate_image_dalle2 - DALL-E 2 Model

Generate images using OpenAI's DALL-E 2 model. Fast and cost-effective option.

Parameters:

• prompt (required): A text description of the desired image (max 1,000 characters)
• output (required): File path where the generated image should be saved
• size (optional): Image size. Options: "256x256", "512x512", "1024x1024". Default: "1024x1024"
• n (optional): Number of images to generate (1-10). Default: 1

Examples

Generate a logo with transparency (GPT-Image-1)

generate_image_gpt(
  prompt="A minimalist geometric logo with circles and triangles",
  output="/path/to/logo.png",
  background="transparent",
  output_format="png",
  quality="high"
)

Generate a cost-efficient image (GPT-Image-1-Mini)

generate_image_gpt_mini(
  prompt="A serene landscape with mountains and a lake at sunset",
  output="/path/to/landscape.png",
  quality="medium",
  size="1024x1024"
)

Generate a high-quality artistic image (DALL-E 3)

generate_image_dalle3(
  prompt="A peaceful zen garden with raked sand and carefully placed stones",
  output="/path/to/zen_garden.png",
  quality="hd",
  style="natural"
)

Generate multiple variations (DALL-E 2)

generate_image_dalle2(
  prompt="A vintage robot reading a newspaper",
  output="/path/to/robot.png",
  n=3,
  size="512x512"
)