Gemini Imagen4 is an MCP server that enables AI image generation using Google's Imagen 4.0 models. It transforms text descriptions into high-quality images through the Model Context Protocol, with outputs saved locally for easy access and management. Key capabilities include: - Three model variants (Standard, Fast, and Ultra) offering different trade-offs between generation speed and image quality - Configurable aspect ratios including 1:1, 3:4, 4:3, 9:16, and 16:9 - Support for PNG and JPEG output formats - Built-in safety filtering with reason reporting for content moderation - Local file storage with timestamped filenames in a designated directory - MCP resource integration for browsing generated images within compatible AI assistants The server requires a Google Gemini API key and can be run locally via npx or installed globally through npm.

🎨 Gemini-Imagen4

Local AI Image Generation MCP Server powered by Google's Imagen 4.0 models

Generate stunning images from text descriptions using Google's cutting-edge Imagen 4.0 models through the Model Context Protocol (MCP). Images are saved locally to ./generated-images/ directory.

✨ Features

🚀 Quick Start

Prerequisites

Installation & Usage

#### Option 1: Install from npm (Recommended)

npm install -g gemini-imagen4

Then run:

gemini-imagen4

#### Option 2: Run directly with npx

npx gemini-imagen4

#### Option 3: Clone and run locally

git clone https://github.com/chug2k/gemini-imagen4.git
cd gemini-imagen4
npm install
npm run dev

You'll be prompted to enter your Gemini API key when you first run it.

Usage

Once running, the server provides:

#### Tool: generate_image_from_text Generate images from text descriptions with optional customization.

Parameters:

Example:

{
  "prompt": "A majestic dragon soaring through a sunset sky",
  "model": "imagen-4.0-ultra-generate-preview-06-06",
  "aspectRatio": "16:9",
  "outputMimeType": "image/png"
}

Response: Images are saved to ./generated-images/ with timestamped filenames like 1754998591_majestic_dragon_soaring.png

#### Resource: generated-images Browse the generated images directory via MCP resources protocol.

🔌 Adding to Claude Desktop

To use this server with Claude Desktop, add it to your MCP configuration:

Edit your claude_desktop_config.json:

{
  "mcpServers": {
    "gemini-imagen4": {
      "command": "npx",
      "args": ["gemini-imagen4"],
      "env": {
        "GEMINI_API_KEY": "your-api-key-here"
      }
    }
  }
}

Replace your-api-key-here with your actual Gemini API key, then restart Claude Desktop.

🔧 Configuration

The server requires a Gemini API key and supports the following configuration:

# smithery.yaml
runtime: typescript
startCommand:
  type: http
  configSchema:
    type: object
    required: ["geminiApiKey"]
    properties:
      geminiApiKey:
        type: string
        title: "Gemini API Key"
        description: "Your Google Gemini API key"
      modelName:
        type: string
        title: "Model Name"
        description: "Default Imagen model to use"
        default: "imagen-4.0-generate-preview-06-06"
        enum: 
          - "imagen-4.0-generate-preview-06-06"
          - "imagen-4.0-fast-generate-preview-06-06"
          - "imagen-4.0-ultra-generate-preview-06-06"

🏃‍♂️ Development

Run the development server:

npm run dev

📦 Model Variants

🛡️ Safety & Content Filtering

All images are processed through Google's built-in safety filters. If content is filtered, the tool will return the reason for filtering instead of an image.

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

📄 License

MIT License - see LICENSE file for details.

🔗 Links

---

Built with ❤️ using Google's Imagen 4.0 and the Model Context Protocol