Skip to main content
skill CX Codex

nano-banana-2 · 101-skills — Claude Code skill: image generation via Gemini

A skill for Claude Code, Codex, Cursor and other platforms that generates images with Google Gemini 3.1 Flash Image Preview (Nano Banana 2) through the belt CLI. Runs on text prompts, edits existing images, and accepts up to 14 input files.

no license 240 installs ★ 31 updated 2026-08-31

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓
npx skills add 101-skills/superpowers --skill nano-banana-2 -a claude-code
Open on GitHub

What nano-banana-2 does

Calls belt app run google/gemini-3-1-flash-image-preview with JSON parameters: prompt (required), images (array of URLs or paths, up to 14 JPEG/PNG/WebP files for editing), num_images, aspect_ratio (1:1, 16:9, 9:16, 4:3, 3:4, auto), resolution (1K, 2K, 4K; defaults to 1K), and enable_google_search to ground generation in live data—weather, news, current facts.

Returns an images array with the output files, a description text field, and output_meta with pricing metadata. Requires belt login and the inference.sh belt CLI installed before first run; commands fail without it. Supports the inference.sh Python SDK with streaming progress updates.

Prompts can specify style (photorealistic, illustration, watercolor, anime, 3D render), composition (close-up, aerial view, portrait), lighting (golden hour, neon, dramatic shadows), and texture, color, mood detail. Generates multiple variants in one call and handles custom aspect ratios.

What it helps with

  • generate an image from a text description with Gemini 3.1 Flash Image
  • edit an existing image based on a prompt
  • get four logo or illustration variants in one request
  • create a 16:9 panoramic scene in 4K resolution
  • ground image generation in live Google Search data

Example prompts for the agent

  • Draw a banana in space, photorealistic, using nano-banana-2
  • Edit landscape.jpg: add a rainbow in the sky
  • Create four minimalist coffee shop logo variants, 1:1 aspect ratio

How to install nano-banana-2

1Skills CLI (Vercel)

npx skills add 101-skills/superpowers --skill nano-banana-2 -a claude-code

2Manually: put the files in place

~/.claude/skills/nano-banana-2/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/nano-banana-2/

Step by step, with commands for each platform: How to install skills in Claude Code: five ways

Questions about nano-banana-2

How to install nano-banana-2 in Claude Code?

Run npx skills add 101-skills/superpowers --skill nano-banana-2 -a claude-code; the skill file lands in ~/.claude/skills/nano-banana-2/SKILL.md.

What does nano-banana-2 do?

Generates images from text and edits existing images via Google Gemini 3.1 Flash Image Preview. Supports up to 14 input files, resolution up to 4K, custom aspect ratios, and grounding in Google Search data.

What does nano-banana-2 require?

The belt CLI from inference.sh and belt login authorization. Without them, belt app run commands will not execute.

Similar skills

All skills for AI and ML →

nano-banana-2 by other authors comes first.

nano-banana-2 prime-skills Skill to quickly generate images via Google Nano Banana 2 on RunComfy. Includes optimized prompting patterns, in-image typography support, and web grounding. CX Codex +1 448,600installs nano-banana-2 intellectronica Drives the Nano Banana 2 API to generate and edit images, with control over resolution and aspect ratio. CX Codex +1 920installs nano-banana-edit prime-skills nano-banana-edit edits images with Google Nano Banana 2 on RunComfy, preserving subject identity, swapping backgrounds, and handling batches up to 20 inputs. CX Codex +1 447,900installs nano-banana-pro-openrouter official github nano-banana-pro-openrouter — a GitHub Copilot skill that calls Gemini 3 Pro Image through OpenRouter to generate and edit pictures. One Python script covers prompt-only creation, single-image edits, and combining up to three inputs. CX Codex +1 9,910installs nano-banana-image acedatacloud nano-banana-image is a skill for generating and editing images through AceDataCloud's API, using text prompts and reference URLs with 1K, 2K, or 4K output. CX Codex +1 232installs nano-banana-pro-prompts-recommend-skill YouMind-OpenLab nano-banana-pro-prompts-recommend-skill is a Claude Code skill that finds image generation prompts in a library of 10,000+ texts. Prompts for Nano Banana Pro also work with DALL-E, Midjourney, Stable Diffusion, and Flux. CX Codex +1 1installs nano-banana-pro intellectronica nano-banana-pro is a Claude Code skill that generates and edits images through Gemini 3 Pro Image API from the terminal. It creates illustrations or modifies existing files. CX Codex +1 3,200installs muapi-nano-banana SamurAIGPT muapi-nano-banana formulates reasoning-driven image prompts in Gemini 3 style and sends them to muapi.ai. Useful when you need legible typography and physically accurate lighting. CX Codex +1 3,000installs

From the same set81

The whole set →
python-sdk python-sdk teaches your coding assistant how to use the inference.sh Python package to run generative apps and assemble agents. CX Codex +1 241installs twitter-thread-creation twitter-thread-creation guides your agent in writing high-engagement X threads, formatting hook tweets, and dispatching posts via the belt CLI. CX Codex +1 240installs pitch-deck-visuals pitch-deck-visuals builds an investor-deck outline with slide design rules and data visuals. Use it for fundraising decks, demo days, grant proposals, or startup presentations. CX Codex +1 240installs ai-rag-pipeline ai-rag-pipeline — a skill for building RAG pipelines: web search supplies the facts, a model writes the cited answer. Handy for research agents and claim checking. CX Codex +1 240installs qwen-image-2 qwen-image-2 is a Claude Code skill for creating and editing images with Alibaba Qwen-Image-2.0. It responds to Qwen Image and image-generation requests. CX Codex +1 240installs qwen-image-2-pro qwen-image-2-pro — a skill for Claude Code and Cursor that generates images via Alibaba Qwen-Image-2.0-Pro. Renders multi-line text in posters and banners. CX Codex +1 239installs linkedin-content linkedin-content — a skill for drafting LinkedIn posts: hook formulas, formatting rules, post types, character limits and the signals the feed rewards. CX Codex +1 239installs widgets-ui widgets-ui — a skill for Claude Code, Cursor, Codex and other agents: builds React/Next.js components from JSON descriptions. The agent outputs a widget structure; the renderer assembles forms, cards and layouts in the project. CX Codex +1 239installs

Project sponsors

Companies that keep this analytics open