image-brother

MCP (Model Context Protocol) server for AI-powered image generation using Duomi Nano Banana API (Google Gemini).

Features

• Text-to-image generation - Create images from text prompts
• Image editing/transformation - Edit and transform images with AI (supports up to 5 reference images)
• Support for multiple models (Gemini 2.5 Pro / Gemini 3 Pro)
• Configurable aspect ratios and resolutions (1K/2K/4K)
• Automatic task polling and result retrieval
• Easy configuration via environment variables

Installation

Via npm

npm install -g image-brother

Via npx (no installation)

npx image-brother

Configuration

Configure your Duomi API key as an environment variable in your MCP client.

Get your API key from: https://duomiapi.com

Usage with Claude Desktop

Add this to your Claude Desktop configuration with your API key:

macOS/Linux

Edit ~/Library/Application Support/Claude/claude_desktop_config.json:

{
  "mcpServers": {
    "image-brother": {
      "command": "npx",
      "args": ["-y", "image-brother"],
      "env": {
        "DUOMI_API_KEY": "your-duomi-api-key-here"
      }
    }
  }
}

Windows

Edit %APPDATA%\Claude\claude_desktop_config.json:

{
  "mcpServers": {
    "image-brother": {
      "command": "npx",
      "args": ["-y", "image-brother"],
      "env": {
        "DUOMI_API_KEY": "your-duomi-api-key-here"
      }
    }
  }
}

Available Tools

1. text_to_image - Text-to-Image Generation

Generate an image from a text prompt.

Parameters

| Parameter | Type | Required | Description | Default | |-----------|------|----------|-------------|---------| | prompt | string | ✅ | Text description of the image to generate | - | | model | string | ❌ | Model: gemini-2.5-pro-image-preview or gemini-3-pro-image-preview | gemini-2.5-pro-image-preview | | aspect_ratio | string | ❌ | Aspect ratio (e.g., 16:9, 1:1, 3:4) | 16:9 | | image_size | string | ❌ | Resolution: 1K, 2K, or 4K (4K only for pro model) | 2K |

Example Usage in Claude

User: Generate an image of a lion on the grassland

Claude: [Calls text_to_image tool with prompt="A lion on the grassland"]

Response Format

{
  "success": true,
  "taskId": "05ed364d-b392-8d18-97c2-eb6ea797541b",
  "imageUrl": "https://cdn3.dmiapi.com/attachments/gemini/.../image.jpeg",
  "fileName": "output.jpeg",
  "description": "Here is an image of a lion on the grassland",
  "prompt": "A lion on the grassland",
  "model": "gemini-2.5-pro-image-preview",
  "aspectRatio": "16:9",
  "imageSize": "2K"
}

2. edit_image - Image Editing/Transformation

Edit or transform images using text prompts and reference images. Upload up to 5 reference images to guide the generation.

Parameters

| Parameter | Type | Required | Description | Default | |-----------|------|----------|-------------|---------| | prompt | string | ✅ | Text description for the image editing/transformation | - | | image_urls | array | ✅ | Array of reference image URLs (1-5 images) | - | | model | string | ❌ | Model: gemini-2.5-pro-image-preview or gemini-3-pro-image-preview | gemini-2.5-pro-image-preview | | aspect_ratio | string | ❌ | Aspect ratio (e.g., 16:9, 1:1, 3:4). Recommended to omit for auto-adaptation | "" (auto-adapt) | | image_size | string | ❌ | Resolution: 1K, 2K, or 4K (4K only for pro model) | 2K |

Example Usage in Claude

User: Make this photo look like a watercolor painting
[User provides image URL: https://example.com/photo.jpg]

Claude: [Calls edit_image tool with:
  prompt="Convert to watercolor painting style",
  image_urls=["https://example.com/photo.jpg"]
]

Response Format

{
  "success": true,
  "taskId": "3fb45a4a-750d-39bb-04da-4337a0f450bd",
  "imageUrl": "https://cdn3.dmiapi.com/attachments/gemini/.../output.jpeg",
  "fileName": "output.jpeg",
  "description": "Watercolor painting style applied",
  "prompt": "Convert to watercolor painting style",
  "referenceImages": ["https://example.com/photo.jpg"],
  "model": "gemini-2.5-pro-image-preview",
  "aspectRatio": "auto-adapt",
  "imageSize": "2K"
}

Notes

• Maximum 5 reference images allowed
• For single reference image, aspect_ratio defaults to auto-adapt to the reference image
• For multiple reference images, you can specify a custom aspect_ratio if needed

Models

Standard Model: gemini-2.5-pro-image-preview

• Resolution: 1K, 2K
• Cost: 0.07 yuan/image
• Good for general use

Pro Model: gemini-3-pro-image-preview

• Resolution: 1K, 2K, 4K
• Cost: 0.14 yuan/image
• Higher quality output

Aspect Ratios

Supported aspect ratios:

• 16:9 - Landscape (default, 1920x1080)
• 1:1 - Square
• 3:4 - Portrait
• 4:3 - Landscape
• 9:16 - Vertical
• And more...

Development

Build from source

git clone <your-repo-url>
cd image-brother
npm install
npm run build

Run locally

npm run build
node dist/index.js

Watch mode

npm run watch

Publishing to npm

# Login to npm
npm login

# Publish
npm publish

Troubleshooting

DUOMI_API_KEY environment variable not set

Make sure you've configured the env section in your MCP client configuration with your API key:

{
  "mcpServers": {
    "image-brother": {
      "command": "npx",
      "args": ["-y", "image-brother"],
      "env": {
        "DUOMI_API_KEY": "your-api-key-here"
      }
    }
  }
}

Task timeout

If image generation takes too long, the server will timeout after 30 polling attempts (60 seconds). Try again with a simpler prompt.

API errors

Check that your API key is valid and you have sufficient credits at https://duomiapi.com.

License

MIT

Credits

• Powered by Duomi API
• Uses Google Gemini Nano Banana image generation
• Built with Model Context Protocol SDK