How this command is triggered — by the user, by Claude, or both
Slash command
/nano-banana:nano-bananaThis command is limited to the following tools:
The summary Claude sees in its command listing — used to decide when to auto-load this command
# /nano-banana - Generate AI Images
You are helping the user generate an image using the nano-banana CLI, powered by Google's Gemini image generation models.
## Tool Location
The CLI tool is at: `${CLAUDE_PLUGIN_ROOT}/tools/cli.ts`
Run with: `bun run "${CLAUDE_PLUGIN_ROOT}/tools/cli.ts" [PROMPT] [OPTIONS]`
## Prerequisites Check
Before first use, verify dependencies are installed:
If `NEEDS_INSTALL`, run:
## Interactive Flow
### 1. Check What's Provided
Parse the command arguments:
- `$ARGUMENTS.prompt` - Image prompt
- `$ARGUMENTS.size` - Size (512, 1K, 2K, 4K)
- `$ARGUMENTS.mo...You are helping the user generate an image using the nano-banana CLI, powered by Google's Gemini image generation models.
The CLI tool is at: ${CLAUDE_PLUGIN_ROOT}/tools/cli.ts
Run with: bun run "${CLAUDE_PLUGIN_ROOT}/tools/cli.ts" [PROMPT] [OPTIONS]
Before first use, verify dependencies are installed:
cd "${CLAUDE_PLUGIN_ROOT}/tools" && [ -d "node_modules" ] && echo "READY" || echo "NEEDS_INSTALL"
If NEEDS_INSTALL, run:
cd "${CLAUDE_PLUGIN_ROOT}/tools" && bun install
Parse the command arguments:
$ARGUMENTS.prompt - Image prompt$ARGUMENTS.size - Size (512, 1K, 2K, 4K)$ARGUMENTS.model - Model (flash, pro)$ARGUMENTS.aspect-ratio - Aspect ratio$ARGUMENTS.output - Output filename$ARGUMENTS.transparent - Transparent background$ARGUMENTS.ref - Reference image path$ARGUMENTS.no-search - Disable search groundingIf prompt is missing, ask the user:
What image would you like to generate? Describe the subject, style, and composition.
If size/model/aspect were not provided, ask with sensible defaults:
Size (if not provided, default to 1K):
Model (if not provided, default to flash):
Aspect Ratio (only ask if relevant to the prompt):
If output is not provided, generate a descriptive default based on the prompt:
[descriptive-slug]-[timestamp]
Use -d flag to place in ~/Downloads or the user's preferred directory.
Construct the full command:
bun run "${CLAUDE_PLUGIN_ROOT}/tools/cli.ts" "USER_PROMPT" \
[-s SIZE] \
[-m MODEL] \
[-a ASPECT_RATIO] \
[-o OUTPUT_NAME] \
[-d OUTPUT_DIR] \
[-r REFERENCE_IMAGE] \
[-t] \
[--no-search]
After completion, report:
/nano-banana "A sunset over mountains" --size 2K --aspect-ratio 16:9 --output sunset
-> Build command and run immediately
/nano-banana "A minimalist logo for a tech startup"
-> Use defaults (1K, flash, 1:1), generate with descriptive filename
/nano-banana
-> Ask for prompt, use sensible defaults, generate
/nano-banana "pixel art robot character" --transparent --output robot
-> Run with -t flag, report transparent PNG output
/nano-banana "make this image brighter and more vibrant" --ref ./screenshot.png
-> Run with -r flag for image editing
| Alias | Model | Cost/1K | Best For |
|---|---|---|---|
flash, nb2 | Gemini 3.1 Flash | ~$0.067 | Speed, volume, iteration (web + image search grounding) |
pro, nb-pro | Gemini 3 Pro | ~$0.134 | Final quality, complex (web search grounding only) |
--costs flag--no-search to disable grounding for abstract/creative promptsnpx claudepluginhub lifegenieai/lifegenie-claude-marketplace --plugin nano-banana/gemini-imageGenerates images from text descriptions using Google Gemini AI. Interactively selects model (e.g., gemini-3.1-flash-image-preview) and aspect ratio (e.g., 1:1). Requires GEMINI_API_KEY and Python 3.
/ai-image-generatorGenerates AI images using Gemini or GPT APIs from a description, with optional purpose like hero, icon, og, or illustration.
/nano-bananaGenerates or edits images using Google's Gemini Image model. Accepts a prompt and optional image path for editing.