Nano Banana Pro

Nanobanana

A Claude Code skill for generating images with Gemini 3 Pro Image (Nano Banana Pro).

The skill automatically transforms your requests into optimized prompts that exploit Nano Banana Pro's best features: perfect text rendering, logical reasoning for infographics, and photorealistic consistency.

Install (plugin marketplace)

/plugin marketplace add 0-to-1-Labs/claude-marketplace
/plugin install nanobanana@0to1-labs

Then set your API key:

1. Get one at Google AI Studio
2. Add to your shell config:

   export GEMINI_API_KEY="your-key-here"
   

3. Restart Claude Code

Alternative: standalone install (no marketplace)

curl -sL https://raw.githubusercontent.com/johnpsasser/nanobanana/main/install.sh | bash

This copies the skill into ~/.claude/skills/nanobanana. Set your API key the same way as above.

Usage

Just ask Claude Code to generate images. The skill activates automatically.

Simple:

Generate an image of a fluffy cat on a cloud

Infographic:

Create a diagram showing how an espresso machine works

Poster with text:

Make a poster for a jazz night called "Blue Moon" on Friday

Portrait:

Portrait of a woman with freckles and green eyes, wearing headphones

Edit an existing image:

Take portrait.png and replace the background with a snowy mountain at golden hour

You can also ask for a specific shape or quality ("make it a 9:16 phone wallpaper", "render it at 4K"), and the skill wires that to the model's aspect ratio and resolution controls. Add "make a quick draft" to use the faster, cheaper Flash model instead of Pro.

Overview

Nano Banana Pro is a reasoning-based model. It does best with natural language descriptions, not keyword soup. This skill handles the translation:

• Detects the type of image you want (infographic, typography, portrait, general)
• Builds a structured prompt with subject, environment, specific text, style, and technical parameters
• Adds photography or illustration details that match the request

The model is particularly good at rendering text in images. If your image needs signage, labels, or titles, just describe what the text should say.

Manual Installation

If you prefer not to run the install script:

1. Create the skill directory:

   mkdir -p ~/.claude/skills/nanobanana/scripts
   

2. Copy the files:

   curl -sL https://raw.githubusercontent.com/johnpsasser/nanobanana/main/SKILL.md \
     -o ~/.claude/skills/nanobanana/SKILL.md
   curl -sL https://raw.githubusercontent.com/johnpsasser/nanobanana/main/requirements.txt \
     -o ~/.claude/skills/nanobanana/requirements.txt
   curl -sL https://raw.githubusercontent.com/johnpsasser/nanobanana/main/scripts/generate.py \
     -o ~/.claude/skills/nanobanana/scripts/generate.py
   

3. Install the dependencies (the script also auto-installs these on first run):

   pip install google-genai pillow
   

4. Set your API key (see above)

5. Restart Claude Code

How It Works

1. You describe what you want
2. The skill analyzes your request and picks a pattern (infographic, typography, character, or general)
3. It builds an enhanced prompt with all the details Nano Banana Pro needs
4. The image generates and saves to your current directory

Images are saved as nanobanana_YYYYMMDD_HHMMSS.png in your current directory.

By default it uses Nano Banana Pro (gemini-3-pro-image) for the best text rendering and fidelity. Quick drafts can use the faster, cheaper Flash model (gemini-3.1-flash-image). All generated images carry an invisible SynthID watermark.

Troubleshooting

"GEMINI_API_KEY not set"

Make sure it's exported in your shell. Check with echo $GEMINI_API_KEY. If it's empty, add it to your .zshrc or .bashrc and restart your terminal.

"google-genai not installed"

The skill tries to auto-install it (and pillow, used for image editing) on first run. If that fails — usually because of multiple Python versions or an externally-managed environment — install manually with one of:

pip install --user google-genai pillow
pip install --break-system-packages google-genai pillow

API errors

Check that your key is valid at Google AI Studio. Free tier has usage limits.

Links

• Gemini API Image Generation Docs
• Nano Banana Pro Documentation
• Nano Banana Hackathon Kit

License

MIT. Do whatever you want with it.