README.md
Google Gemini image generation MCP server
@saroby/nanobanana-mcp
Google Gemini image generation MCP server (Nano Banana).
Features
- generate_image - Generate images using Google Gemini models
flashmode: gemini-2.5-flash-image (fast, ~2-3s)promode: gemini-3-pro-image-preview (high quality, ~5-8s)
- list_images - List generated images in a directory
Setup
Get API Key
- Go to Google AI Studio
- Create an API key
Install in Claude Code
claude mcp add nanobanana -e GEMINI_API_KEY=your-key-here -- npx -y @saroby/nanobanana-mcp
Manual Usage
GEMINI_API_KEY=your-key-here npx @saroby/nanobanana-mcp
Tools
generate_image
| Parameter | Type | Required | Description |
|---|---|---|---|
prompt |
string | Yes | Image description (1-8192 chars) |
model |
"flash" | "pro" |
No | Model selection (default: flash) |
aspect_ratio |
enum | No | 1:1, 16:9, 9:16, 4:3, 3:4 |
negative_prompt |
string | No | Elements to exclude |
output_dir |
string | No | Save directory (default: ./nanobanana-images) |
count |
number | No | Number of images 1-4 (default: 1) |
list_images
| Parameter | Type | Required | Description |
|---|---|---|---|
directory |
string | No | Directory to search (default: ./nanobanana-images) |
Development
npm install
npm run build
npm run dev # watch mode
License
MIT
Tools (2)
generate_imageGenerate images using Google Gemini models with support for custom aspect ratios and negative prompts.list_imagesList generated images in a directory.Environment Variables
GEMINI_API_KEYrequiredAPI key from Google AI StudioConfiguration
claude_desktop_config.json
{"mcpServers": {"nanobanana": {"command": "npx", "args": ["-y", "@saroby/nanobanana-mcp"], "env": {"GEMINI_API_KEY": "your-key-here"}}}}Try it
→Generate a high-quality image of a futuristic city skyline using the pro model.
→Create 4 images of a cute robot reading a book with a 16:9 aspect ratio.
→Generate an image of a mountain landscape, but make sure to exclude any clouds using a negative prompt.
→List all the images I have generated in the default directory.