MCPInstall

nano-banana-prompts

Generate optimized prompts for Gemini 2.5 Flash Image (Nano Banana). Use for image generation, crafting photo prompts, art styles, or multi-turn editing workflows with best practices.

Installs
607
Last 24h
—
Trending rank
—
Security
Not audited

Install

Works with every agent the skills CLI supports.

npx skills add https://github.com/secondsky/claude-skills --skill nano-banana-prompts
  1. 01

    Install

    Run the command in your project. The CLI asks which agents to install for.

  2. 02

    Describe the task

    Ask your agent for something this skill covers. It loads nano-banana-prompts automatically when relevant.

  3. 03

    Review the result

    The agent follows the SKILL.md steps, scripts and references bundled with the skill.

SKILL.md

The instructions your agent reads when the skill is triggered.

Nano Banana Image Prompt Generator

Generate high-quality prompts for Gemini 2.5 Flash Image using proven best practices from October 2025.

Core Principle

Describe scenes naturally, don't list keywords.

❌ woman, business suit, office, professional ✅ A professional businesswoman in navy suit stands confidently in modern glass-walled office overlooking city skyline. Warm afternoon light streams through windows.

Prompt Structure

Basic Template

A [SHOT TYPE] of [SUBJECT] [ACTION] in [ENVIRONMENT]. 
[LIGHTING] creates [MOOD]. [CAMERA/STYLE DETAILS]. [ASPECT RATIO].

Components

  1. Subject & Action - Who/what and what they're doing
  2. Setting - Where the scene takes place
  3. Lighting - Natural/artificial light sources and quality
  4. Mood - Emotional tone or atmosphere
  5. Technical Details - Camera, lens, style specifics
  6. Format - Aspect ratio or dimensions

Prompt Length

  • Short (10-20 words): Quick generation, minimal control
  • Medium (50-100 words): ⭐ RECOMMENDED - balanced control
  • Long (100-200 words): Maximum control for complex scenes

Photography Prompts

Template

Photorealistic [SHOT TYPE] of [SUBJECT], [ACTION], in [ENVIRONMENT]. 
Illuminated by [LIGHTING], creating [MOOD]. Captured with [CAMERA/LENS], 
emphasizing [DETAILS]. [ASPECT RATIO] format.

Camera Control

Lenses:

  • 35mm, 50mm (standard)
  • 85mm portrait (close-ups)
  • 10-24mm wide (landscapes)
  • 60-105mm macro (details)

Lighting:

  • Golden hour (warm, soft)
  • Blue hour (cool, atmospheric)
  • Three-point (studio)
  • Dramatic side (high contrast)
  • Rembrandt (classic portrait)

Settings:

  • Shallow DOF: f/1.4-2.8 (blurred background)
  • Deep focus: f/11-22 (everything sharp)
  • Motion blur for action

Example

Professional corporate headshot of mid-30s woman with shoulder-length brown hair 
in charcoal blazer. She's smiling warmly at camera against soft gray backdrop. 
Shot with 85mm f/2.0, studio three-point lighting creates gentle shadows defining 
facial features. Ultra-high resolution, LinkedIn-ready quality. 1:1 format.

Art & Illustration Prompts

Template

[STYLE] illustration of [SUBJECT] [ACTION/POSE]. [COMPOSITION] with 
[COLOR PALETTE]. [TECHNIQUE/MEDIUM]. Inspired by [REFERENCE], 
emphasizing [QUALITIES].

Common Styles

  • Watercolor painting
  • Digital art
  • Oil painting
  • Pencil sketch
  • Ink drawing
  • Anime/manga
  • Pixel art
  • Minimalist vector

Example

Watercolor illustration of young girl reading book under oak tree. Seated 
cross-legged in sun-dappled clearing with wildflowers. Soft, loose brushstrokes 
with pastel color palette of greens, yellows, pinks. Transparent washes create 
dreamy, ethereal quality. Inspired by children's book illustration, emphasizing 
wonder and tranquility. 4:5 portrait format.

Text Rendering

Best Practices:

  • Keep under 25 characters total
  • Maximum 2-3 phrases
  • Specify font style (bold, script, sans-serif)
  • Put exact text in quotes
  • Describe position explicitly

Example:

Movie poster with bold red text "SUMMER" at top and "Adventure Awaits" 
in white script font at bottom. Ocean sunset background. Cinematic style.

Multi-Turn Editing Workflow

Nano Banana excels at conversational editing. Make one change at a time.

Example sequence:

  1. "Create professional headshot of woman in business attire"
  2. "Change background to modern office"
  3. "Make her wearing glasses"
  4. "Add warmer lighting"

Product Photography

Template

High-resolution product photo of [PRODUCT] on [SURFACE]. [LIGHTING] 
to [PURPOSE]. Camera angle: [TYPE] to showcase [FEATURE]. Ultra-realistic, 
sharp focus on [DETAIL]. [ASPECT RATIO].

Example

High-resolution product photo of stainless steel smartwatch on white marble 
surface. Soft diffused studio lighting from 45-degree angle to minimize 
reflections and show screen clearly. Camera angle: slight overhead to showcase 
circular face and metal band texture. Ultra-realistic, sharp focus on watch 
face display. 1:1 format for e-commerce.

Landscape Photography

Template

[TYPE] landscape of [LOCATION] at [TIME]. [WEATHER/ATMOSPHERE]. 
[FOREGROUND ELEMENTS]. Shot with [LENS] from [VANTAGE POINT]. 
[COMPOSITION]. Professional landscape photography with [GRADING]. [MOOD].

Example

Dramatic landscape of mountain lake at golden hour. Wispy clouds catch 
warm orange light over snow-capped peaks. Weathered dock extends into 
mirror-still water as foreground. Shot with 16mm wide-angle from low 
lakeside vantage. Rule-of-thirds composition places horizon in upper third. 
Professional landscape photography with warm color grading. Serene yet majestic.

Abstract Art

Template

[MOVEMENT] abstract exploring [CONCEPT]. [COMPOSITION] with [COLORS/TEXTURES]. 
[MEDIUM] emphasizing [GOALS]. [CONTEXT].

Example

Bold geometric abstract exploring tension and harmony. Overlapping angular 
shapes with vibrant primary colors against deep black. Digital art emphasizing 
sharp edges and color interaction. Large-scale wall art format, 16:9.

Common Pitfalls

Avoid:

  • Keyword lists without context
  • Vague descriptions
  • Conflicting style instructions
  • Overly complex text rendering
  • Too many simultaneous changes in edits

Instead:

  • Use complete sentences
  • Be specific about details
  • Maintain style consistency
  • Keep text simple
  • Edit iteratively

Style References

When requesting specific artistic styles, reference:

  • Art movements (impressionism, art deco)
  • Artists (in style of Van Gogh's brushwork)
  • Media (watercolor, oil painting, digital art)
  • Eras (1950s photography, modern minimalist)

Aspect Ratios

  • 1:1 - Square (social media, profile pics)
  • 16:9 - Landscape (presentations, headers)
  • 9:16 - Portrait (mobile, stories)
  • 4:5 - Portrait (Instagram posts)
  • 2:3 - Classic portrait
  • 3:2 - Standard photo

Quick Prompt Checklist

✅ Subject clearly described ✅ Action or pose specified ✅ Environment/setting detailed ✅ Lighting source and quality ✅ Mood or atmosphere ✅ Style or technical specs ✅ Aspect ratio chosen ✅ Complete sentences, not keywords

Common Issues & Solutions

Prompt Quality Issues

Issue: Generic or vague output Cause: Prompt too short or lacks specificity Solution: Expand to 50-100 words. Add lighting details, mood, and specific technical specs (lens, style, etc.).

Example:

  • ❌ Too vague: "A woman in an office"
  • ✅ Specific: "Professional businesswoman in navy suit stands confidently in modern glass-walled office overlooking city skyline at golden hour. Warm afternoon light streams through floor-to-ceiling windows creating soft rim lighting. Shot with 50mm f/2.0, shallow depth of field blurs cityscape background. Confident yet approachable mood. 4:5 portrait format."

Issue: Conflicting style instructions Cause: Mixing incompatible styles (e.g., "photorealistic watercolor") Solution: Choose one primary style. For photography, use camera terminology. For art, use medium/technique terminology.

Issue: Text rendering failures Cause: Too much text, complex fonts, or unclear positioning Solution:

  • Limit to under 25 characters total
  • Use simple, clear fonts (bold, script, sans-serif)
  • Specify exact position ("at top", "centered", "bottom right")
  • Put text in quotes: "SUMMER"

Issue: Unwanted elements in image Cause: Ambiguous phrasing or conflicting descriptors Solution: Be explicit about what should NOT appear. Use precise, unambiguous language.

Generation Issues

Issue: Image doesn't match prompt Cause: Prompt too complex or contradictory elements Solution:

  1. Simplify prompt to core concept
  2. Use multi-turn editing workflow
  3. Start basic, refine iteratively

Issue: Inconsistent results across generations Cause: Vague descriptors allowing wide interpretation Solution: Add specific details about colors, materials, lighting angles, camera settings, and composition rules (rule of thirds, golden ratio).

Issue: Poor composition Cause: No compositional guidance in prompt Solution: Explicitly mention composition rules:

  • "Rule of thirds composition"
  • "Centered subject with symmetrical framing"
  • "Leading lines draw eye to focal point"
  • "Foreground, mid-ground, background layers"

Multi-Turn Editing Issues

Issue: Too many changes at once causing confusion Cause: Requesting multiple edits simultaneously Solution: Make one change per turn. Wait for result before next edit.

Example sequence:

  1. ✅ "Change background to sunset"
  2. ✅ "Make the lighting warmer"
  3. ✅ "Add reflections on water" vs.
  4. ❌ "Change background to sunset with warmer lighting and add reflections"

Issue: Losing context from earlier turns Cause: Vague references without reinforcing key details Solution: Briefly reinforce context: "In this mountain landscape, add a hiking trail in the foreground"

Prompt Validation

Before submitting a prompt, verify:

Content Completeness

  • Subject clearly identified (who/what)
  • Action or state described (what subject is doing)
  • Environment/setting specified (where)
  • Lighting described (source, quality, direction)
  • Mood or atmosphere indicated (emotional tone)
  • Style or medium specified (photography, watercolor, etc.)
  • Aspect ratio chosen (1:1, 16:9, 4:5, etc.)

Technical Specifications

For Photography:

  • Camera/lens mentioned (50mm, 85mm, wide-angle)
  • Camera settings if relevant (f/2.8, shallow DOF)
  • Lighting type (golden hour, studio three-point, dramatic side)
  • Shot type (close-up, wide shot, overhead)

For Art/Illustration:

  • Medium specified (watercolor, digital art, oil painting)
  • Technique mentioned (loose brushstrokes, sharp vector)
  • Color palette described (pastels, vibrant, monochrome)
  • Style reference if applicable (children's book, manga, minimalist)

Quality Checks

  • Uses complete sentences, not keyword lists
  • Specific rather than vague ("charcoal blazer" vs. "suit")
  • No conflicting style instructions
  • Text under 25 characters if included
  • Length appropriate: 50-100 words for balanced control
  • No contradictory elements (photorealistic watercolor)

Text Rendering Validation

If prompt includes text:

  • Total characters under 25
  • Maximum 2-3 text phrases
  • Text in quotes ("SUMMER")
  • Font style specified (bold, script, sans-serif)
  • Position explicitly stated (top, bottom, centered)
  • Text color contrasts with background

Testing Your Prompts

Quick Test

Generate image and evaluate:

  1. Subject accuracy: Is the main subject as described?
  2. Composition: Does layout match prompt intent?
  3. Lighting: Is lighting source and quality correct?
  4. Style: Does aesthetic match requested style?
  5. Technical specs: Are camera/art specs honored?
  6. Mood: Does image convey intended atmosphere?

Refinement Workflow

If image doesn't match expectations:

Step 1: Identify the gap

  • What element is wrong/missing?
  • Is it subject, lighting, style, composition, or mood?

Step 2: Use multi-turn editing

  • Don't regenerate from scratch
  • Make targeted edit: "Make the lighting softer"
  • One change at a time

Step 3: Refine prompt for future use

  • If you need to regenerate, add more specificity to weak areas
  • Save successful prompts as templates

Comparison Test

For critical images, generate 2-3 variations with slightly different prompts:

Version A: Medium detail (50-75 words) Version B: High detail (100-150 words) Version C: Different style/lighting approach

Compare results to find optimal prompt verbosity and approach.

Example: Complete Validation

Prompt:

Photorealistic close-up portrait of elderly fisherman with weathered hands mending
fishing net at dawn. Seated on wooden dock with misty harbor in background. Soft
golden light from rising sun creates warm rim lighting on his profile. Shot with
85mm f/2.0 creating shallow depth of field, net and hands in sharp focus while
background gently blurs. Contemplative and peaceful mood. 4:5 portrait format.

Validation:

  • ✅ Subject: elderly fisherman
  • ✅ Action: mending fishing net
  • ✅ Environment: wooden dock at harbor
  • ✅ Lighting: golden hour, rim lighting
  • ✅ Mood: contemplative, peaceful
  • ✅ Technical: 85mm f/2.0, shallow DOF
  • ✅ Aspect ratio: 4:5 portrait
  • ✅ Complete sentences: yes
  • ✅ Specific details: weathered hands, wooden dock, misty harbor
  • ✅ Length: ~85 words (optimal range)
  • ✅ Style consistency: photorealistic throughout

Expected result: High-quality, detailed image matching all specifications.

Troubleshooting Guide

ProblemLikely CauseFix
Blurry/low qualityMissing technical specsAdd camera, lens, "ultra-high resolution"
Wrong colorsVague color descriptionSpecify exact colors, hex codes if needed
Bad compositionNo compositional guidanceAdd "rule of thirds", "centered", etc.
Missing elementsElement mentioned too latePut critical elements early in prompt
Generic lookKeyword listingRewrite as natural sentences
Text garbledToo much textReduce to <25 chars, simplify font
Wrong styleConflicting instructionsChoose one consistent style
Unexpected objectsAmbiguous phrasingBe more explicit and specific

FAQ

What does the nano-banana-prompts skill do?

Generate optimized prompts for Gemini 2.5 Flash Image (Nano Banana). Use for image generation, crafting photo prompts, art styles, or multi-turn editing workflows with best practices.

How do I install nano-banana-prompts?

Run npx skills add https://github.com/secondsky/claude-skills --skill nano-banana-prompts in your project. The skills CLI detects the agents you use and copies the skill into the right folder. For Claude Code you can also place the skill folder at ~/.claude/skills/nano-banana-prompts/ (personal) or .claude/skills/nano-banana-prompts/ (project).

Which AI agents can use nano-banana-prompts?

Any agent that supports the open Agent Skills format (a folder with a SKILL.md file), including Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, Windsurf, Cline, OpenCode and more.

Is nano-banana-prompts safe to install?

No partner security audit is available yet. Skills can run commands and scripts, so read the SKILL.md and any bundled files before installing.

Who maintains nano-banana-prompts?

It is published by secondsky in secondsky/claude-skills. It has 607 recorded installs.

More from secondsky/claude-skills

View all →

tailwind-v4-shadcn

secondsky/claude-skills

Production-tested setup for Tailwind CSS v4 with shadcn/ui, Vite, and React. Use when: initializing React projects with Tailwind v4, setting up shadcn/ui, implementing dark mode, debugging CSS variable issues, fixing theme switching, migrating from Tailwind v3, or encountering color/theming problems. Covers: @theme inline pattern, CSS variable architecture, dark mode with ThemeProvider, component composition, vite.config setup, common v4 gotchas, and production-tested patterns.

Frontend & Design8.1K installs

aceternity-ui

secondsky/claude-skills

100+ animated React components (Aceternity UI) for Next.js with Tailwind. Use for hero sections, parallax, 3D effects, or encountering animation, shadcn CLI integration errors.

Frontend & Design1.5K installs

vitest-testing

secondsky/claude-skills

Modern TypeScript/JavaScript testing with Vitest. Fast unit and integration tests, native ESM support, Vite-powered HMR, and comprehensive mocking. Use for testing TS/JS projects.

Testing & QA1.2K installs

motion

secondsky/claude-skills

Motion (Framer Motion) React animation library. Use for drag-and-drop, scroll animations, gestures, SVG morphing, or encountering bundle size, complex transitions, spring physics errors.

Frontend & Design1.1K installs

playwright

secondsky/claude-skills

Browser automation and E2E testing with Playwright. Auto-detects dev servers, writes clean test scripts. Test pages, fill forms, take screenshots, check responsive design, validate UX, test login flows, check links, automate any browser task. Use for cross-browser testing, visual regression, API testing, component testing in TypeScript/JavaScript and Python projects.

Testing & QA1K installs

api-testing

secondsky/claude-skills

HTTP API testing for TypeScript (Supertest) and Python (httpx, pytest). Test REST APIs, GraphQL, request/response validation, authentication, and error handling.

Backend & APIs892 installs

Popular ai & agents skills

View category →

google-agents-cli-publish

google/agents-cli

This skill should be used when the user wants to "publish an agent", "publish my ADK agent", "register an agent with Gemini", "publish to Gemini" (formerly Gemini Enterprise), or needs guidance on the agents-cli publish gemini command. Also use when the user wants to "manage agents in Agent Registry", "list/update/delete registered agents", or "register an MCP server". Covers ADK vs A2A registration modes, programmatic and interactive usage, flag reference, auto-detection from deployment metadata, Agent Registry fleet management, and troubleshooting. Part of the agents-cli skills suite. Do NOT use for deployment (use google-agents-cli-deploy).

AI & Agents513.9K installs

google-agents-cli-eval

google/agents-cli

This skill should be used when the user wants to "run an evaluation", "evaluate my agent", "evaluate my ADK agent", "write an eval dataset", "analyze eval failures", "compare eval results", "optimize agent", or needs guidance on the Gemini platform eval methodology and the Quality Flywheel. Covers eval metrics, dataset schema, LLM-as-judge scoring, and common failure causes. Applies to any agents-cli project, whatever framework the agent is written in. Do NOT use for agent API code patterns (ADK: use google-agents-cli-adk-code), deployment (use google-agents-cli-deploy), or project scaffolding (use google-agents-cli-scaffold).

AI & Agents513.5K installs

skill-creator

anthropics/skills

Create new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.

OfficialAI & Agents416.4K installs

codex-pet

prime-skills/runcomfy-agent-skills

Codex Pet generator on RunComfy. Build a Codex-compatible Codex Pet spritesheet.webp + pet.json from a single reference image, drop it into ${CODEXHOME:-$HOME/.codex}/pets/ / and Codex picks it up as a custom Codex Pet next to the 8 built-ins. This skill produces the exact Codex Pet atlas Codex expects (1536x1872 PNG/WebP, 8 cols x 9 rows, 192x208 cells, 9 animation states — idle, running-right, running-left, waving, jumping, failed, waiting, running, review). Calls OpenAI GPT Image 2 edit ONCE via the local RunComfy CLI as runcomfy run openai/gpt-image-2/edit to produce a canonical Codex Pet pose, then assembles all 9 animation rows programmatically with ImageMagick micro-transforms — no Codex Pro, no $imagegen, no OPENAIAPIKEY required, only RUNCOMFYTOKEN. Triggers on "codex pet", "create codex pet", "make codex pet", "hatch codex pet", "/hatch image", "desktop pet codex", "codex pets", "spritesheet.webp", or any explicit ask to build a custom pet for OpenAI Codex.

AI & Agents402.1K installs

full-output-enforcement

leonxlnx/taste-skill

Overrides default LLM truncation behavior. Enforces complete code generation, bans placeholder patterns, and handles token-limit splits cleanly. Apply to any task requiring exhaustive, unabridged output.

AI & Agents401K installs

executing-plans

obra/superpowers

Use when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available

AI & Agents245.7K installs