Reasoning-driven image generation using structured creative

Muapi nano banana

by SamurAIGPT · MIT license · ★ 4,317 on GitHub · v0.1.0 · updated Sep 8

Reasoning-driven image generation using structured creative briefs (Gemini 3 style) — generates high-fidelity images via muapi.ai with logic-based prompting

---
name: muapi-nano-banana
version: 0.1.0
description: Reasoning-driven image generation using structured creative briefs (Gemini 3 style) — generates high-fidelity images via muapi.ai with logic-based prompting
---

# 🍌 Nano-Banana Expert Skill (Gemini 3 Style)

**A specialized skill for AI Agents to leverage "Reasoning-Driven" image generation.**
Based on the advanced prompting architecture of Google's Gemini 3 (Nano Banana Pro), this skill moves beyond keyword stuffing to structured, logic-based creative briefs.

## Core Competencies

1. **Reasoning-Driven Prompting**: Using natural language logic to define physics, lighting, and spatial relationships.
2. **Structured Creative Briefs**: Implementing the "Perfect Prompt" formula: `Subject + Action + Context + Composition + Lighting`.
3. **Text Rendering Precision**: Explicitly defining typography and signifiers for legible text integration.
4. **Contextual Grounding**: Using "Search Grounding" logic (simulated) to anchor generations in real-world accuracy.

Overview

A specialized skill for AI Agents to leverage "Reasoning-Driven" image generation. Based on the advanced prompting architecture of Google's Gemini 3 (Nano Banana Pro), this skill moves beyond keyword stuffing to structured, logic-based creative briefs.

The SKILL.md below is the copy AgentAlley read and checked. The repo has changed it since; the other files are from commit 5519622.

Manual install (npx, ZIP)

Claude Codenpx skills add SamurAIGPT/Generative-Media-Skills --skill muapi-nano-banana -g -a claude-code

Claude Code (manual)

mkdir -p ~/.claude/skills && curl -sL https://codeload.github.com/SamurAIGPT/Generative-Media-Skills/tar.gz/5519622e885a \
  | tar -xz -C ~/.claude/skills --strip-components=3 Generative-Media-Skills-5519622e885a/.opencode/skills/muapi-nano-banana

Windows (PowerShell)

New-Item -ItemType Directory -Force "$HOME\.claude\skills" | Out-Null; Invoke-WebRequest "https://m.agentalley.io/zip/6271ff325ba8fdc87b54/muapi-nano-banana.zip" -OutFile "$env:TEMP\muapi-nano-banana.zip"; Expand-Archive "$env:TEMP\muapi-nano-banana.zip" -DestinationPath "$HOME\.claude\skills" -Force

Claude.ai / desktopDownload the ZIP and upload it under Settings → Capabilities → Skills.

Codex, Cursor, other agentsPoint them at SKILL.md. The other files are plain text.

Connects with Agent Alley MCP · agentalley.io/mcp

Files included

SKILL.md3.7 KB

Read the files here

SamurAIGPT/55196221 file

SKILL.md3.7 KB

Shared references1SKILL.md reads these from elsewhere in the repo. Not in the ZIP.

generate-nano-art.shlibrary/visual/nano-banana/scripts/1.8 KB

Get them on GitHub ↗

SKILL.md

Show the full text80 lines

muapi-nano-banana/SKILL.md80 lines · 3.7 KB

PreviewCode

Outline

RawView on GitHub

🍌 Nano-Banana Expert Skill (Gemini 3 Style)

A specialized skill for AI Agents to leverage "Reasoning-Driven" image generation. Based on the advanced prompting architecture of Google's Gemini 3 (Nano Banana Pro), this skill moves beyond keyword stuffing to structured, logic-based creative briefs.

Core Competencies

Reasoning-Driven Prompting: Using natural language logic to define physics, lighting, and spatial relationships.
Structured Creative Briefs: Implementing the "Perfect Prompt" formula: Subject + Action + Context + Composition + Lighting.
Text Rendering Precision: Explicitly defining typography and signifiers for legible text integration.
Contextual Grounding: Using "Search Grounding" logic (simulated) to anchor generations in real-world accuracy.

🏗️ Technical Specification

1. The "Perfect Prompt" Formula

Component	Description	Example
Subject	Detailed entity description	"A stoic robot barista with exposed copper wiring"
Action	Dynamic interaction	"Pouring a latte art leaf with mechanical precision"
Context	Environment & Atmosphere	"Inside a neon-lit cyberpunk cafe at midnight"
Composition	Camera & Lens choice	"Close-up, 85mm lens, f/1.8 aperture"
Lighting	Mood & Direction	"Volumetric blue rim light, warm cafe glow"
Style	Aesthetic anchor	"Cinematic, photorealistic, 4K production value"

2. Advanced Features

Negative Constraint Logic: Instead of "no blurry," use "Ensure sharp focus on the subject's eyes."
Identity Consistency: (Simulated) "Maintain consistent facial structure across variations."
Text Integration: Use double quotes for specific text: The sign reads "OPEN 24/7".

🧠 Prompt Optimization Protocol (Agent Instruction)

Before calling the script, the Agent MUST rewrite the user's prompt into a logic-driven Reasoning Brief:

NO KEYWORD SOUP: Remove "8k, masterpiece, ultra-detailed." Use full, descriptive sentences.
PHYSICAL CONSISTENCY: Describe how elements interact (e.g., "The light from the crystal shards casts caustic patterns across the obsidian floor").
TEXT PRECISION: If the user wants text, define it precisely: featuring a sign that says "STORE NAME" in a weathered serif font.
OPTICAL DIRECTIVES: Specify lens behavior: Shallow Depth of Field (f/1.8), Macro Lens, Anamorphic Flare.

🚀 Protocol: Using Nano-Banana

Step 1: Define the Creative Logic

Provide the agent with a subject and a specific scenario.

Step 2: Invoke the Script

The generate-nano-art.sh script translates the logic into a structured Gemini 3-style prompt.

# Generating a reasoning-driven image
bash scripts/generate-nano-art.sh \
  --subject "a glass chess piece" \
  --action "shattering into liquid shards" \
  --context "on a obsidian table" \
  --style "macro photography"

⚠️ Constraints & Guardrails

No Keyword Soup: MANDATORY - Do not use "trending on artstation, masterpiece, 8k". Use natural language descriptions.
Physics Logic: Ensure the prompt describes physically possible lighting and reflection interactions.
Full Sentences: The model parses relationships; use "light reflecting off the water" instead of "water, reflection".

⚙️ Implementation Details

This skill applies a "Logic Wrapper" around the core/media/generate-image.sh primitive, converting fragmented inputs into a coherent, reasoning-ready narrative prompt.

1	---
2	name: muapi-nano-banana
3	version: 0.1.0
4	description: Reasoning-driven image generation using structured creative briefs (Gemini 3 style) — generates high-fidelity images via muapi.ai with logic-based prompting
5	---
6
7	# 🍌 Nano-Banana Expert Skill (Gemini 3 Style)
8
9	A specialized skill for AI Agents to leverage "Reasoning-Driven" image generation.
10	Based on the advanced prompting architecture of Google's Gemini 3 (Nano Banana Pro), this skill moves beyond keyword stuffing to structured, logic-based creative briefs.
11
12	## Core Competencies
13
14	1. Reasoning-Driven Prompting: Using natural language logic to define physics, lighting, and spatial relationships.
15	2. Structured Creative Briefs: Implementing the "Perfect Prompt" formula: `Subject + Action + Context + Composition + Lighting`.
16	3. Text Rendering Precision: Explicitly defining typography and signifiers for legible text integration.
17	4. Contextual Grounding: Using "Search Grounding" logic (simulated) to anchor generations in real-world accuracy.
18
19	---
20
21	## 🏗️ Technical Specification
22
23	### 1. The "Perfect Prompt" Formula
24
25	\| Component \| Description \| Example \|
26	\| :--- \| :--- \| :--- \|
27	\| Subject \| Detailed entity description \| "A stoic robot barista with exposed copper wiring" \|
28	\| Action \| Dynamic interaction \| "Pouring a latte art leaf with mechanical precision" \|
29	\| Context \| Environment & Atmosphere \| "Inside a neon-lit cyberpunk cafe at midnight" \|
30	\| Composition \| Camera & Lens choice \| "Close-up, 85mm lens, f/1.8 aperture" \|
31	\| Lighting \| Mood & Direction \| "Volumetric blue rim light, warm cafe glow" \|
32	\| Style \| Aesthetic anchor \| "Cinematic, photorealistic, 4K production value" \|
33
34	### 2. Advanced Features
35	- Negative Constraint Logic: Instead of "no blurry," use "Ensure sharp focus on the subject's eyes."
36	- Identity Consistency: (Simulated) "Maintain consistent facial structure across variations."
37	- Text Integration: Use double quotes for specific text: `The sign reads "OPEN 24/7"`.
38
39	---
40
41	## 🧠 Prompt Optimization Protocol (Agent Instruction)
42
43	Before calling the script, the Agent MUST rewrite the user's prompt into a logic-driven Reasoning Brief:
44
45	1. NO KEYWORD SOUP: Remove "8k, masterpiece, ultra-detailed." Use full, descriptive sentences.
46	2. PHYSICAL CONSISTENCY: Describe how elements interact (e.g., "The light from the crystal shards casts caustic patterns across the obsidian floor").
47	3. TEXT PRECISION: If the user wants text, define it precisely: `featuring a sign that says "STORE NAME" in a weathered serif font`.
48	4. OPTICAL DIRECTIVES: Specify lens behavior: Shallow Depth of Field (f/1.8), Macro Lens, Anamorphic Flare.
49
50	---
51
52	## 🚀 Protocol: Using Nano-Banana
53
54	### Step 1: Define the Creative Logic
55	Provide the agent with a subject and a specific scenario.
56
57	### Step 2: Invoke the Script
58	The `generate-nano-art.sh` script translates the logic into a structured Gemini 3-style prompt.
59
60	```bash
61	# Generating a reasoning-driven image
62	bash scripts/generate-nano-art.sh \
63	--subject "a glass chess piece" \
64	--action "shattering into liquid shards" \
65	--context "on a obsidian table" \
66	--style "macro photography"
67	```
68
69	---
70
71	## ⚠️ Constraints & Guardrails
72
73	- No Keyword Soup: MANDATORY - Do not use "trending on artstation, masterpiece, 8k". Use natural language descriptions.
74	- Physics Logic: Ensure the prompt describes physically possible lighting and reflection interactions.
75	- Full Sentences: The model parses relationships; use "light reflecting off the water" instead of "water, reflection".
76
77	---
78
79	## ⚙️ Implementation Details
80	This skill applies a "Logic Wrapper" around the `core/media/generate-image.sh` primitive, converting fragmented inputs into a coherent, reasoning-ready narrative prompt.