Gemini Image Skill

🍌 gemini-image-skill

Equip Claude Code with Nano Banana — give Claude the power to generate and edit images.

This is a [Claude Code](https://docs.anthropic.com/en/docs/claude-code) skill that integrates Google Gemini's Nano Banana image models, enabling Claude to generate images from text, edit existing images, iteratively refine images through multi-turn conversation, and output in 2K/4K resolution.

Capabilities

Feature	Description
Text-to-Image	Generate images from text descriptions
Image Editing	Modify existing images with natural language instructions
Multi-turn Iteration	Refine images through conversation until you're satisfied
HD Output	Support 2K / 4K resolution via Pro model

Setup

1. Get a Gemini API Key

Go to [Google AI Studio](https://aistudio.google.com/apikey) to create your API key.

**Important**: Nano Banana models do **NOT** support the free tier. You **must** bind a credit card and upgrade to a paid plan to use the image generation API.

2. Set Environment Variable

Add your API key to `~/.zshrc` or `~/.bashrc`:

export GEMINI_API_KEY="your-api-key-here"

3. Install Python Dependencies

pip install google-genai Pillow

4. Install the Skill

Clone this repo into Claude Code's skills directory:

git clone https://github.com/haotongl/gemini-image-skill.git ~/.claude/skills/gemini-image

Restart Claude Code and you're good to go.

Usage

Once installed, just talk to Claude Code in natural language:

> Draw a cyberpunk cat sitting on the moon

> Change the background of this image to a starry sky  (drag in an image or specify a path)

> Make the colors brighter, add a glow effect  (multi-turn iteration)

> Generate the final version in 2K HD

Models

Model	Codename	Use Case
gemini-3.1-flash-image-preview	Nano Banana 2	Fast generation/editing, default