banana-claude
AI image generation skill for Claude Code - Creative Director powered by Gemini
Install / Use
npx skills add AgriciDaniel/banana-claudeInstalls into whichever agent you are using.
SKILL.md
Installable skill definition
Quality Score
Category
OtherSupported Platforms
Our assessment of banana-claude
banana-claude scores 89/100 on our quality scale, 105th of 301 Other skills we index (top 35%).
Its SKILL.md is 8.9 KB long, well organised into 21 sections with 9 code examples: a thorough specification that gives an agent plenty to work with.
With 1,183 GitHub stars, it is one of the more widely adopted skills in the catalogue.
Maintenance, license and trust
- The repository was last updated 4 days ago, so banana-claude is actively maintained.
- Our last check on 2026-10-18 found the source still online.
- It is released under the MIT license, a permissive license that allows use, modification and commercial use with attribution.
- Its trust signals score 95/100, with 1 caution from licensing, adoption, age or documentation. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.
Safety scan
No issues foundOur scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands (1 minor note below). An AI review of the same text found nothing harmful.
- noteInstalls by piping a downloaded script into a shellline 65
curl -fsSL https://raw.githubusercontent.com/AgriciDaniel/banana-claude/main/install.sh | bash
AI review by kimi-k2.7-code on 2026-09-24. Automated pattern scan on 2026-09-24. It catches known dangerous patterns, not every risk — read a skill before letting an agent act on it.
banana-claude compared with similar skills
All 4 of these similar skills score higher than banana-claude; compare them before choosing.
| Skill | Score | Stars | Updated | Format |
|---|---|---|---|---|
| banana-claude (this skill)by AgriciDaniel | 89 | 1.2k | 4d ago | SKILL.md |
| taste-skillby Leonxlnx | 100 | 110.6k | today | CLAUDE.md |
| LocalAIby mudler | 100 | 50.2k | today | MCP Server |
| algorithmic-artby anthropics | 100 | 177.9k | 44d ago | SKILL.md |
| pptxby anthropics | 100 | 177.9k | 44d ago | SKILL.md |
Frequently asked questions
- How do I install banana-claude?
- Run
npx skills add AgriciDaniel/banana-claude. The install tabs above show the steps for each supported agent. - Which AI agents does banana-claude work with?
- It is written for Claude Code and Gemini CLI, as a SKILL.md file. Other agents that read the same format can often use it too.
- Is banana-claude safe to use?
- Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands (1 minor note below). An AI review of the same text found nothing harmful. It is MIT-licensed and scores 95/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
- Is banana-claude still maintained?
- The repository was last updated 4 days ago, so banana-claude is actively maintained.
Skill content
View source on GitHub
Banana Claude
AI image generation skill for Claude Code where Claude acts as Creative Director using Google's Gemini Nano Banana models.
Unlike simple API wrappers, Claude interprets your intent, selects domain expertise, constructs optimized prompts using Google's official 5-component formula, and orchestrates Gemini for the best possible results.
<details> <summary>Table of Contents</summary>
- Installation
- Quick Start
- Commands
- How It Works
- What Makes This Different
- The 5-Component Prompt Formula
- Domain Modes
- Models
- Architecture
- Requirements
- Changelog
- Contributing
- License
Installation
Plugin Install (Recommended)
Add the marketplace and install:
/plugin marketplace add AgriciDaniel/banana-claude
/plugin install banana-claude@banana-claude-marketplace
Or test locally:
git clone --depth 1 https://github.com/AgriciDaniel/banana-claude.git
claude --plugin-dir ./banana-claude
<details>
<summary>Standalone Install (without plugin system)</summary>
git clone --depth 1 https://github.com/AgriciDaniel/banana-claude.git
bash banana-claude/install.sh
One-liner (curl):
curl -fsSL https://raw.githubusercontent.com/AgriciDaniel/banana-claude/main/install.sh | bash
With MCP Setup:
git clone --depth 1 https://github.com/AgriciDaniel/banana-claude.git
cd banana-claude
./install.sh --with-mcp YOUR_API_KEY
</details>
Get a free API key at Google AI Studio.
Quick Start
# Start Claude Code
claude
# Generate an image
/banana generate "a hero image for a coffee shop website"
# Edit an existing image
/banana edit ~/photo.png "remove the background"
# Multi-turn creative session
/banana chat
# Browse 2,500+ prompt database
/banana inspire
Claude will ask about your brand, select the right domain mode (Cinema, Product, Portrait, Editorial, UI, Logo, Landscape, Infographic, Abstract), construct a detailed prompt with lighting and composition, set the right aspect ratio, and generate.

Commands
| Command | Description |
|---------|-------------|
| /banana | Interactive -- Claude detects intent and guides you |
| /banana generate <idea> | Full Creative Director pipeline |
| /banana edit <path> <instructions> | Intelligent image editing |
| /banana chat | Multi-turn visual session (maintains consistency) |
| /banana inspire [category] | Browse 2,500+ prompt database |
| /banana batch <idea> [N] | Generate N variations (default: 3) |
| /banana setup | Configure MCP and API key |
| /banana preset [list\|create\|show\|delete] | Manage brand/style presets |
| /banana cost [summary\|today\|estimate] | View cost tracking and estimates |
How It Works

What Makes This Different
- Intent Analysis -- Understands what you actually need (blog header? app icon? product shot?)
- Domain Expertise -- Selects the right creative lens (Cinema, Product, Portrait, Editorial, UI, Logo, Landscape, Infographic, Abstract)
- 5-Component Prompt Formula -- Constructs prompts with Subject + Action + Location/Context + Composition + Style (includes lighting)
- Prompt Adaptation -- Translates patterns from a 2,500+ curated prompt database to Gemini's natural language format
- Post-Processing -- Crops, removes backgrounds, converts formats, resizes for platforms
- Batch Variations -- Generates N variations rotating different components
- Session Consistency -- Maintains character/style across multi-turn conversations
- 4K Resolution Output -- Up to 4096×4096 with
imageSizecontrol - 14 Aspect Ratios -- Including ultra-wide 21:9 for cinematic compositions
The 5-Component Prompt Formula

Instead of sending "a cat in space" to Gemini, Claude constructs:
A medium shot of a tabby cat floating weightlessly inside the cupola module of the International Space Station, paws outstretched toward a floating droplet of water, Earth visible through the circular windows behind. Soft directional light from the windows illuminates the cat's fur with a blue-white rim light, while the interior has warm amber instrument panel glow. Captured with a Canon EOS R5, 35mm f/2.0 lens, slight barrel distortion emphasizing the curved module interior. In the style of a National Geographic cover story on the ISS, with the sharp documentary clarity of NASA mission photography.
Components used: Subject (tabby cat, physical detail) → Action (floating, paw gesture) → Location/Context (ISS cupola, Earth visible) → Composition (medium shot, curved framing) → Style (Canon R5, National Geographic documentary, directional window light + amber instruments)
Domain Modes

| Mode | Best For | Example | |------|----------|---------| | Cinema | Dramatic, storytelling | "A noir detective scene in a rain-soaked alley" | | Product | E-commerce, packshots | "Photograph my handmade candle for Etsy" | | Portrait | People, characters | "A cyberpunk character portrait for my game" | | Editorial | Fashion, lifestyle | "Vogue-style fashion shot for my brand" | | UI/Web | Icons, illustrations | "A set of onboarding illustrations" | | Logo | Branding, identity | "A minimalist logo for a tech startup" | | Landscape | Backgrounds, wallpapers | "A misty mountain sunrise for my desktop" | | Infographic | Data, diagrams | "Visualize our Q1 sales growth" | | Abstract | Generative art, textures | "Voronoi tessellation in neon gradients" |
Models
| Model | ID | Notes |
|-------|----|-------|
| Flash 3.1 (default) | gemini-3.1-flash-image-preview | Fastest, newest, 14 aspect ratios, up to 4K |
| Flash 2.5 | gemini-2.5-flash-image | Stable fallback |
Architecture
banana-claude/ # Claude Code Plugin
├── .claude-plugin/
│ ├── plugin.json # Plugin manifest
│ └── marketplace.json # Marketplace catalog
├── skills/banana/ # Main skill
│ ├── SKILL.md # Creative Director orchestration (v1.4)
│ ├── references/
│ │ ├── prompt-engineering.md # 5-component formula, banned keywords, safety rephrase
│ │ ├── gemini-models.md # Model specs, rate limits, capabilities
│ │ ├── mcp-tools.md # MCP tool parameters and responses
│ │ ├── post-processing.md # ImageMagick/FFmpeg pipelines, green screen
│ │ ├── cost-tracking.md # Pricing table, usage guide
│ │ └── presets.md # Brand preset schema and examples
│ └── scripts/
│ ├── setup_mcp.py # Configure MCP in Claude Code
│ ├── validate_setup.py # Verify installation
│ ├── generate.py # Direct API fallback -- generation
│ ├── edit.py # Direct API fallback -- editing
│ ├── cost_tracker.py # Cost logging and summaries
│ ├── presets.py # Brand/style preset management
│ └── batch.py # CSV batch workflow parser
└── agents/
└── brief-constructor.md # Subagent for prompt construction
Requirements
- Claude Code
- Node.js 18+ (for npx)
- Google AI API key (free tier: ~5-15 RPM / ~20-500 RPD, cut ~92% Dec 2025)
- ImageMagick (optional, for post-processing)
Uninstall
Plugin:
/plugin uninstall banana-claude@banana-claude-marketplace
Standalone:
bash banana-claude/install.sh --uninstall
Contributing
Contributions welcome! Please open an issue or submit a pull request.
License
MIT License -- see LICENSE for details.
Built for Claude Code by @AgriciDaniel
Author
Built by Agrici Daniel - AI Workflow Architect.
- Blog - Deep dives on AI marketing automation
- AI Marketing Hub - Free community, 2,800+ members
- YouTube - Tutorials and demos
- All open-source tools
Related Skills
taste-skill
110.6kAgent skills that give coding agents taste: art direction, reference images and a browser check instead of generic AI pages
LocalAI
50.2kLocalAI is the open-source AI engine. Run any model - LLMs, vision, voice, image, video - on any hardware. No GPU required.
algorithmic-art
177.9kCreating algorithmic art using p5.js with seeded randomness and interactive parameter exploration. Use this when users request creating art using code, generative art, algorithmic art, flow fields, or particle systems.
pptx
177.9kUse this skill any time a .pptx or .potx file is involved in any way — as input, output, or both. This includes: creating slide decks, pitch decks, or presentations; reading, parsing, or extracting text from any .pptx or .potx file (even if the extracted content will be used elsewhere, like in an em…
Languages
Trust signals
From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.
