Store Screenshots

3. Capture screenshots for each key screen:

mkdir -p screenshots/ios
xcrun simctl io booted screenshot screenshots/ios/01_home.png

4. Ask developer to navigate to each screen, capture 4-6 screenshots.

5. For Android (if emulator running):

mkdir -p screenshots/android
adb shell screencap -p /sdcard/screen.png && adb pull /sdcard/screen.png screenshots/android/01_home.png

Step 2b: AI Generation (nano-banana-mcp)

1. Read app description from app.json or metadata files.
2. For each screen, use generate_image tool:

Prompt template:

iPhone 16 Pro Max screenshot of a {app_category} app showing {screen_description}.
Modern iOS UI, clean minimal design, {color_scheme} theme.
Full screen app UI, no device frame, 1290x2796 resolution.

3. Generate 4-6 images. Save to screenshots/ios/ and screenshots/android/.

Step 2c: Process Only

Skip generation, go directly to Step 3.

Step 3: Post-Processing

3-1. Create config if missing

If screenshots/config.json doesn't exist, generate it from app metadata:

{
  "texts": {
    "en-US": ["Track Your Calories", "Easy Food Log", "Beautiful Charts", "Set Goals"],
    "ko": ["칼로리 추적", "간편한 음식 기록", "아름다운 차트", "목표 설정"]
  },
  "fontSize": 56,
  "fontColor": "#FFFFFF",
  "overlayHeight": 200,
  "font": null
}

Write marketing-oriented text based on actual app features.

3-2. Run processor

python3 ${CLAUDE_SKILL_DIR}/../../scripts/process_screenshots.py --project . --platform both

This resizes to exact dimensions (iOS 1290×2796, Android 1080×1920) and adds text overlays.

Step 4: Upload

iOS:

fastlane ios upload_screenshots

Android:

fastlane android upload_screenshots

Step 5: Report

Screenshots Complete
====================
Processed: {count} images
iOS:     fastlane/screenshots/{langs}/
Android: fastlane/metadata/android/{langs}/images/phoneScreenshots/
Uploaded: {yes/no}