gemini-cli-extensions/nanobanana — reverse-engineered prompt

Build me a Gemini CLI extension called Nano Banana that lets me create and edit images from the command line using Google Gemini image models.

I want simple slash commands for generating images from a prompt, editing an existing image with plain English instructions, restoring old or damaged photos, making app icons and favicons, creating seamless patterns, making visual story sequences, and creating diagrams or flowcharts. It should also have one natural language command where I can just say what I need and it figures out the right action.

Please make it save images with friendly filenames, avoid overwriting files, support multiple variations and styles, and optionally open a preview after generation. It should use an API key from an environment variable and let me choose the image model from an environment variable, with the newest Nano Banana model as the default.

Use TypeScript and make it feel polished and reliable. Look up the current Gemini CLI and Gemini image model docs online if you need to.