Start from words
Turn a prompt into production-ready artwork with explicit size and aspect-ratio controls.
Generate, edit, and art-direct images with Google Gemini models—without breaking your creative flow.
v2.1.0 · Node.js 22+ · MIT licensed · built for humans and agents
$ npm install -g @crapthings/nb-cli
nb-cli turns Gemini's image models into predictable terminal commands for new visuals, precise edits, and clean app icons.
Turn a prompt into production-ready artwork with explicit size and aspect-ratio controls.
Bring an existing image and describe the exact change in natural language.
Generate a focused 1:1 icon with prompt guidance tailored to simple visual identities.
Choose the model that fits the moment. Nano Banana 2.1 is the default, with 1K output and support for 2K and 4K. Use Lite for fast 1K exploration, or select nano-banana-2 for 512 drafts.
# Generate a cinematic wide image nb generate "a glass pavilion in a pine forest" \ --size 2k --aspect-ratio 16:9 \ --output pavilion.jpg # Change only the atmosphere nb edit pavilion.jpg \ "replace daylight with blue hour" \ --output pavilion-blue.jpg # Fast 1K concept nb generate "playful banana sticker" --model lite
From app icons to ultrawide hero art, choose a useful resolution and a composition that already fits its destination.
The bundled Codex skill teaches agents how to frame prompts, choose image parameters, and use nb-cli inside a project workflow.