What it does

wardrobe-builder is a developer-centric asset management system for personal fashion. It transforms messy closet photos, shopping screenshots, and multi-item montages into a clean, structured JSON database. It handles the heavy lifting of OCR brand detection, visual feature extraction (color, fit, material), and automated image processing like background removal or intelligent cropping.

How it works

This skill provides a robust pipeline consisting of a directory scaffolder, an ingestion engine, a safe-write layer, and a linter. It can detect garment boundaries, split composite images into individual items, and map them to a strict schema. To ensure data integrity, it uses a confidence-based review queue—questionable data is flagged for your approval rather than polluting your database.

Supported Tools

• Python-based automation for local file management
• OCR & Computer Vision for attribute tagging
• Lark/Feishu CLI integration for remote photo uploading
• JSON-based local storage (no cloud lock-in)

Why use this skill

Manually tagging a 200-item wardrobe is tedious. Prompting an AI to "remember my clothes" results in hallucinations and lost context. This skill provides a permanent, local "source of truth" that other agents (like stylists) can query reliably. It includes a built-in linter to fix broken image paths and schema violations, keeping your data production-ready.