---

name: media-intelligence-v1 description: Image and video registry management with automated slot injection

Media Intelligence Skill v1.0

Purpose: Manage the asset registry, video ingestion, and image transparency fixes.

1) Non-Negotiables (Policy)

Asset Organization

Copyclient/src/assets/
├── products/           # Product images (True Alpha PNGs)
├── backgrounds/        # Video backgrounds
├── icons/              # UI icons
└── images/             # General images

public/
└── videos/             # Hosted video files

Image Standards

• Product images: True Alpha (transparent backgrounds)
• No baked checkerboard artifacts
• Preferred format: PNG with transparency
• Use rembg for background removal when needed

Video Standards

• Registry: client/src/lib/video-registry.ts
• All videos must be registered
• Generate poster images for preload

---

2) Inputs & Outputs

Video Registration

Copyinterface VideoEntry {
  id: string;
  src: string;
  poster?: string;
  keywords: string[];
  category: 'product' | 'nature' | 'science' | 'abstract';
}

Scripts Available

---

3) Quality Gates

Before Adding Media

• Asset placed in correct directory
• Registered in appropriate registry
• Transparent background (for products)
• Optimized file size

After Adding Media

Copynpm run validate:assets

---

Background Removal Workflow

Using rembg (Python)

Copypython scripts/remove_bg.py input.png output.png

Batch Processing

Copy# Process all product images
for f in client/src/assets/products/*.png; do
  python scripts/remove_bg.py "$f" "${f%.png}-clean.png"
done

---

Video Ingestion Workflow

1. Add video to public/videos/
2. Run indexer: tsx scripts/index-videos.ts
3. Generate poster: tsx scripts/generate-video-posters.ts
4. Update registry with keywords