Product Reel Prep

Inspiration

I create e-commerce content on TikTok for an Algerian audience. Every week my phone fills with product photos and clips, and I lose hours deciding which clips are postable and writing captions in Arabic and French. I wanted a private assistant that does this where my media already lives: on the NAS.

What it does

product-reel-prep turns a messy folder of raw product photos and phone clips into a short-video content kit:

• a clip review per product (reel-ready, trim, needs-reshoot) based on orientation, length and duplicates
• a product fact sheet read from the photos and packaging labels (vision + OCR)
• a shot list of missing shots
• multilingual captions with hashtags, using only facts visible in the media
• a saved content-kit note and optional file tags (after approval)

How I built it

As a pure orchestration SKILL.md built with the official UGOS Skill Creation Tool (uliya-skill-kit 0.7.3). It delegates to file-agent (ug-filesystem, ug-filemgr), photo-agent (ug-photo), document-agent (ug-document) and knowledge-agent (ug-knowledge), and handles every delegation state. It passes the official static checker with zero errors and zero warnings.

Challenges

Preventing invented product facts. Captions are written only from the verified fact sheet, and missing facts are flagged as ask the seller.

What's next

Testing on a real AI NAS: phone-video media info, OCR on packaging, and Arabic caption quality.

Built With

• markdown
• ugreen