Google Image Gen Plugin for Claude Code

A Claude Code plugin for generating images using Google's Gemini API.

What This Plugin Does

This plugin enables Claude Code to generate images using Google's Gemini API. When you ask Claude to create, generate, or edit images, it will use this plugin automatically.

Features:

• Generate images from text prompts
• Edit existing images with AI
• Use reference images for style consistency
• Support for multiple aspect ratios
• Batch generation of multiple variations

Requirements

• uv package manager
• Google AI API key from Google AI Studio

Installation

1. Add the Marketplace & Install

In Claude Code, run:

/plugin marketplace add ypfaff/google-image-gen-plugin
/plugin install google-image-gen@google-image-gen

2. Set Your API Key

Option A: Config file (recommended)

mkdir -p ~/.config/google-image-gen
echo "GOOGLE_AI_API_KEY=your_actual_api_key_here" > ~/.config/google-image-gen/.env

Option B: Environment variable

export GOOGLE_AI_API_KEY=your_actual_api_key_here
# Add to your shell profile (~/.zshrc, ~/.bashrc) for persistence

3. Done

Claude Code will automatically detect and use the plugin when you ask it to generate images.

Usage Examples

Just ask Claude Code naturally:

Simple Image Generation

Generate an image of a red sports car on a mountain road and save it as car.png

Create a minimalist logo for a coffee shop, save as logo.png with 1:1 aspect ratio

With Specific Aspect Ratios

Generate a YouTube thumbnail showing a coding tutorial, use 16:9 aspect ratio

Create a phone wallpaper with abstract geometric shapes, aspect ratio 9:16

Editing Existing Images

Edit the image input.png and change the background to a sunset sky

Using Reference Images

Generate an image in the same style as reference.png but with a cat instead

Multiple Variations

Generate 3 variations of app icons: a rocket, a star, and a lightning bolt

With Style Templates

You can create custom style templates as markdown files with a ## Prompt Template section:

## Prompt Template

\```
Minimalist flat design illustration of {subject}.
Clean lines, bold colors, white background.
Modern corporate style, suitable for presentations.
\```

Then ask Claude:

Generate an image of a laptop using the style template my-style.md

Supported Aspect Ratios

Ratio	Use Case
1:1	Social media posts, profile pictures
3:4	Portrait photos
4:3	Standard landscape
4:5	Instagram portrait
5:4	Large format prints
9:16	Phone wallpapers, Stories, TikTok
16:9	YouTube thumbnails, presentations (default)
21:9	Ultrawide displays, cinematic

Troubleshooting

Error	Cause	Solution
GOOGLE_AI_API_KEY not found	Missing API key	Create ~/.config/google-image-gen/.env with your API key
Rate limit exceeded	Too many API requests	Wait a few minutes or upgrade to paid tier
uv: command not found	uv not installed	Install via curl -LsSf https://astral.sh/uv/install.sh | sh

Development

Local Testing

claude --plugin-dir .

Versioning & Updates

Bump version in plugin.json (semver) and push to main. Claude Code pulls the latest commit from the default branch — git tags are currently ignored. Optionally tag for your own release tracking:

git tag 1.1.0 && git push origin 1.1.0

Users receive updates via /plugin update or automatic update checks.

Architecture Overview

Component	Role
skills/…/SKILL.md	Loaded into Claude's context as skill prompt
main.py, scripts/	Executed via Bash by Claude
${CLAUDE_PLUGIN_ROOT}	Plugin cache path (where plugin files live)
--cwd / working dir	User's project path (where output files are saved)

Credits

This plugin is based on the work from AI Engineer Skool's Google Image Gen API Starter.

For more infos, check out the corresponding video guide.

License

MIT