nano-banana-image-gen

Nano Banana image generation

Procedure

1. Finalize the image generation prompt from the user's request
2. For multimodal instructions (when there is an input image such as a frame image), --image add the argument
3. Run the following command

Text only:

python .claude/skills/nano-banana-image-gen/scripts/nano_banana_gen.py "プロンプトテキスト"

Image + text:

python .claude/skills/nano-banana-image-gen/scripts/nano_banana_gen.py "プロンプトテキスト" --image "画像パス"

Multiple images + text:

python .claude/skills/nano-banana-image-gen/scripts/nano_banana_gen.py "プロンプトテキスト" --image "画像1" --image "画像2"

1. Check the output path and report it to the user

Settings

Project root's .env Set the following environment variables in:

GEMINI_API_KEY=your_api_key_here
GENERATED_IMAGE_OUTPUT_PATH=images/generated

Security: .env If placing in the project root,settings.json の denyRead に .env add and configure so that AI cannot read it.

// .claude/settings.json
{
  "permissions": {
    "deny": ["Read(.env)", "Read(.env.*)"]
  }
}

Variable name	Description
GEMINI_API_KEY	Google Gemini API key (required)
GENERATED_IMAGE_OUTPUT_PATH	Output directory (default: images/generated）

Output

• File name: YYYYMMDD_HHMMSS.png
• Save location: Project root's images/generated/

Dependency package check

For initial errors, install with the following:

pip install google-genai pillow python-dotenv