Claude Banana workflow
Claude Code subagent that turns a rough image idea into a detailed prompt using a seven-part formula for image generators.
- 110 ★ on the repository
- Last commit
- MIT
Minimal access
Why this level
- The agent writes prompt text. Generation scripts use a Google AI key and spend its quota.
Install
- Works in
Manual install
git clone https://github.com/Hainrixz/claude-banana.gitThen change into the folder and run `claude`.
Access and checks
How we reviewMinimal accessThe tool runs nothing and makes no network calls.
- The agent writes prompt text. Generation scripts use a Google AI key and spend its quota.
This is third-party code. Review the repository files before installing.
What it does
The project adds a `prompt-architect` subagent and a prompt knowledge base to Claude Code. You describe a picture in plain words, the agent asks a couple of questions about mood and use, then returns a ready prompt. Every prompt is built from seven parts: subject, style, environment, lighting, action, camera and texture. The agent works in nine modes: cinema, product, portrait, fashion, UI, logo, landscape, abstract and infographic. The base holds over 70 creative techniques, 25 prompt templates and brand presets with colors and mood. Prompts from Midjourney, DALL-E and Stable Diffusion are rewritten for Nano Banana Pro. Optional Python scripts create, edit and batch images through a Google AI key.
Who it is for. Designers, marketers and anyone making images in Nano Banana Pro who is tired of weak results from short descriptions.
Good fit when
- A phrase like a cozy cafe with a cat on books needs to become a detailed prompt with lighting and angle.
- A brand needs a series of images in one palette through a preset.
- Old Midjourney prompts have to be rewritten for Nano Banana Pro.
- One image needs a batch of variations from a list of values.
Not a fit when
- You need an image generator itself; the project writes prompts and creates images only with your Google key.
- You do not work in Claude Code, since the subagent lives in `.claude/agents/`.
- The image is finished and needs manual retouching.
Example request
Help me write a prompt for a cozy coffee shop with warm light and a cat sleeping on a stack of books.Limitations
The agent loads when you run `claude` inside the cloned project folder. The README has no separate plugin install. Prompts are tuned for Google's Nano Banana Pro and fit other generators less well. Direct generation through `scripts/generate.py`, `scripts/edit.py` and `scripts/batch.py` needs a Google AI key and spends its quota, and setup is checked with the `scripts/validate_setup.py` script. Results depend on the image model, so even a detailed prompt does not guarantee the picture you want. The README and agent prompts are in English, and a Spanish README exists.
How to disable. Delete the cloned project folder. Revoke the Google AI key on the keys page if you created it only for this project.
FAQ
How do I set up Claude Banana?
Run `git clone https://github.com/Hainrixz/claude-banana.git`, change into the folder and start `claude`. Claude picks up the `prompt-architect` agent and the project's knowledge base.
What does a prompt consist of?
Seven parts: subject, style, environment, lighting, action, camera and texture. The agent fills them in from your answers to its questions.
Which modes does the agent have?
Nine: cinema, product, portrait, fashion, UI, logo, landscape, abstract and infographic. The mode follows what you want to get.
Do I need a key to write prompts?
No. The agent writes prompts without external keys. A Google AI key is needed only for the scripts that create and edit the images themselves.
How do I create an image from the terminal?
Check setup with `python3 scripts/validate_setup.py`, then run `python3 scripts/generate.py --prompt "your prompt"`. Use `scripts/edit.py` for edits and `scripts/batch.py` for series.
Discussion
No questions or reviews yet.
Related
WorkflowsA set of skills for Claude Code and Codex that assemble detailed image prompts from an element library and composition schemes
Nano Banana Pro Prompts Recommend
AI Image Prompt Recommender
A skill that picks ready image-generation prompts from a large library, shows sample images and assembles a prompt for your text
The skill turns paper text, methods and captions into structured prompts for scientific figures, tuned to the chosen image model.
A skill that searches a library of 10,000+ image prompts, shows sample images and tailors the chosen prompt to your article or script
Over 1,400 popular prompts for NanoBanana and GPT Image from X posts ranked by likes, plus a system prompt that turns a short idea into a detailed one
Nano Banana Pro Prompts by YouMind
awesome-nano-banana-pro-prompts
A large prompt library for Nano Banana Pro: by the author's count more than 10 thousand prompts with examples, sources and a web gallery
README in short
The project teaches Claude Code to turn simple descriptions into professional prompts, while scripts cover running generation and image edits.