Nano Banana App Store Screenshot Skill
Обзор
An agent skill for building premium, consistent App Store screenshot campaigns with Nano Banana Pro. This skill helps an agent: - analyze an app codebase for brand, claims, and differentiators - decide which app screens should be captured - write a layout plan before image generation - lock a style anchor for consistency across the set - generate advertisement-first App Store slides - normalize exports to App Store-safe sizes - build a contact sheet for review The skill is built around a specific sequence so the screenshot set stays consistent and App Store-safe: 1. Scan the codebase - inspect fonts, colors, onboarding copy, pricing language, differentiators, and supported claims - identify the app’s strongest user benefit and the real features worth selling 2. Decide the campaign story - turn the app into a 5 to 7 slide ad narrative - define what each slide should sell in one second 3.
README
Nano Banana ASO
An agent skill for building premium, consistent App Store screenshot campaigns with Nano Banana Pro.
This skill helps an agent:
- analyze an app codebase for brand, claims, and differentiators
- decide which app screens should be captured
- write a layout plan before image generation
- lock a style anchor for consistency across the set
- generate advertisement-first App Store slides
- normalize exports to App Store-safe sizes
- build a contact sheet for review
Workflow
The skill is built around a specific sequence so the screenshot set stays consistent and App Store-safe:
-
Scan the codebase
- inspect fonts, colors, onboarding copy, pricing language, differentiators, and supported claims
- identify the app’s strongest user benefit and the real features worth selling
-
Decide the campaign story
- turn the app into a 5 to 7 slide ad narrative
- define what each slide should sell in one second
-
Ask for specific screenshots
- tell the user exactly which screens to capture
- specify what should be visible in each raw screenshot
-
Write a layout plan
- lock campaign positioning, story spine, consistency spine, crop-safe rules, and export target
- use the plan as the source of truth before any image generation starts
-
Convert copy into render-safe prompts
- keep `` line breaks in planning docs only
- translate final prompt copy into explicit multi-line headline blocks
- explicitly tell the image model not to render literal `` characters
- shorten text-heavy slides before pushing copy toward crop edges
-
Generate a style anchor
- create slide 1 first with Nano Banana Pro
- use it to define typography, background language, panel treatment, device treatment, and overall visual family
- use it to test cropping, font treatment, CTA cues, and text density before the rest of the set is generated
-
Generate the rest of the set
- create slides 2 to N as consistent variations of the approved anchor
- reuse the findings from perfecting slide 1 so the later slides inherit proven spacing and typography rules
- keep the screenshot or product visual as the main focus
- add relevant supporting imagery when it strengthens the message
- explicitly lock follow-up slides to the anchor’s headline font family / type style
- avoid accidental serif drift or giant text-card layouts unless the anchor already uses them
-
Normalize and crop-check exports
- save raw outputs to
native/ - normalize to a final App Store export folder like
export-1320x2868/ - visually inspect the normalized images, not just the raw renders
- verify that no text, screenshot edge, or focal visual gets cropped out
- if cropping is off, regenerate, normalize again, and keep checking until it works
- use stricter center-safe rules on text-heavy slides, not just the default edge margins
- save raw outputs to
-
Package review assets
- generate a
prompt-log.md - generate a
contact-sheet.png - present the normalized final set for review
- generate a
Included
SKILL.mdagents/openai.yamlreferences/scripts/normalize_app_store_exports.pyscripts/build_contact_sheet.py
What It Optimizes For
- screenshots as ads, not tutorials
- consistent campaign-wide visual language
- screenshot-led compositions with relevant supporting imagery
- crop-safe layouts that survive final App Store normalization
- a
native/plusexport-/workflow - render-safe headline formatting for image models
- follow-up slides that keep the anchor’s typography voice
Default Export Pattern
output/
nano-banana-layout-plan.md
prompt-log.md
native/
export-1320x2868/
contact-sheet.png
Model Guidance
- Use
Nano Banana Proonly by default. - Before generation, check
GEMINI_API_KEYfirst, thenGOOGLE_API_KEY. - If neither key is available, ask the user for the Gemini API key before generation.
- Do not switch to another image model as a fallback.
Install
skills.sh
Install it directly with skills.sh:
npx skills add chawlaaditya/NanoBananaASO
Direct clone
If you want to install it manually into a local Codex skills directory, clone it directly:
git clone https://github.com/chawlaaditya/NanoBananaASO.git "${CODEX_HOME:-$HOME/.codex}/skills/nano-banana-app-store-campaign"
Codex built-in installer
If you already have the Codex system skills available locally, you can install from GitHub with the built-in installer:
python3 "${CODEX_HOME:-$HOME/.codex}/skills/.system/skill-installer/scripts/install-skill-from-github.py" \
--repo chawlaaditya/NanoBananaASO \
--path .
After installing, restart Codex so the new skill is picked up.
Use
Example prompt:
Use $nano-banana-app-store-campaign to analyze this app, tell me which screenshots to capture, lock a Nano Banana Pro style anchor, and generate a consistent App Store screenshot campaign.
Examples
Here are a few real campaign contact sheets generated with the workflow:
GentleCal
RunPal
Рекомендуемые инструменты
Попробуйте другой запрос или уберите фильтр.
Установка
npx skillfish add chawlaaditya/nanobananaaso