图像生成与编辑MCP服务

Gemini MCP - Image Generation Server

MCP (Model Context Protocol) server for image generation and editing using Google's Gemini API and ImageMagick.

Quick Install for AI Assistants

AI Assistant Instructions: Follow these steps to install this MCP server for the user. You will need to ask the user for their Gemini API key.

Step 1: Check Prerequisites

# Check Python version (needs 3.10+)
python3 --version

# Check if ImageMagick is installed
magick --version || convert --version

# Check if uv is installed
uv --version

If ImageMagick is missing:

• Arch/CachyOS: sudo pacman -S imagemagick
• Ubuntu/Debian: sudo apt install imagemagick
• macOS: brew install imagemagick

If uv is missing:

curl -LsSf https://astral.sh/uv/install.sh | sh

Step 2: Clone and Install

# Clone the repository
git clone https://github.com/tarpediem/geminigenerator.git
cd geminigenerator

# Create virtual environment and install
uv venv
uv pip install -e .

Step 3: Get the Installation Path

# Get the absolute path (you'll need this)
pwd

Step 4: Ask User for API Key

AI: Ask the user: "Please provide your Gemini API key. You can get one free at https://aistudio.google.com/apikey"

Step 5: Configure Claude Code

The MCP configuration file location depends on the system:

• Linux: ~/.config/claude-code/settings.json
• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json

Add or merge this configuration (replace INSTALLATION_PATH and YOUR_API_KEY):

{
  "mcpServers": {
    "gemini-image": {
      "command": "uv",
      "args": [
        "--directory",
        "INSTALLATION_PATH",
        "run",
        "python",
        "-m",
        "gemini_mcp.server"
      ],
      "env": {
        "GEMINI_API_KEY": "YOUR_API_KEY"
      }
    }
  }
}

Step 6: Restart Claude Code

Tell the user to restart Claude Code to load the new MCP server.

Step 7: Verify Installation

After restart, the following tools should be available:

• gemini_generate_image
• gemini_edit_image
• gemini_generate_with_references
• gemini_describe_image
• im_resize, im_crop, im_rotate, im_flip, im_convert, im_effects, im_composite, im_thumbnail, im_info, im_border

Features

Gemini AI Tools

ImageMagick Tools (Non-AI, Local Processing)

Available Models

Supported Parameters

Aspect Ratios

1:1, 2:3, 3:2, 3:4, 4:3, 4:5, 5:4, 9:16, 16:9, 21:9

Resolutions

• 1K - 1024px (default, fastest)
• 2K - 2048px
• 4K - 4096px (gemini-3-pro only)

Usage Examples

Generate an image

Generate a cyberpunk cityscape at sunset, neon lights reflecting on wet streets

Edit an existing image

Edit /path/to/image.png: add a rainbow in the sky

Generate with style reference

Generate a portrait in the style of the reference images: [/path/to/style1.png, /path/to/style2.png]

Batch processing with ImageMagick

Resize all images in /photos to 800px width, convert to webp, add 5px white border

Manual Installation

git clone https://github.com/tarpediem/geminigenerator.git
cd geminigenerator
uv venv && uv pip install -e .
cp .env.example .env
# Edit .env and add your GEMINI_API_KEY

Environment Variables

Troubleshooting

"GEMINI_API_KEY not set"

Ensure the API key is set in the MCP configuration's env block.

"ImageMagick not found"

Install ImageMagick for your system (see prerequisites).

"Module not found"

Run uv pip install -e . in the project directory.

License

MIT

Credits

Built with FastMCP, google-genai, and Wand.