AuraLoop VTO: Agentic Skin AI & Virtual Try-On Concierge

Inspiration

What it does

How we built it

Challenges we ran into

Accomplishments that we're proud of

What we learned

What's next for AuraLoop VTO: Agentic Skin AI & Virtual Try-On Concierge

Inspiration

Beauty and fashion shoppers get generic recommendations that ignore their actual skin condition, undertone and safety constraints. AuraLoop VTO turns the YouCam API into an agentic, privacy-first try-on concierge.

What it does

• 14-metric YouCam Skin AI diagnostics feeding a contraindication-safe skincare regimen engine (barrier safety gate).
• CIEDE2000 (CIELAB) perceptual shade and apparel-harmony matching instead of naive RGB distance.
• Chained Makeup + Clothes virtual try-on with a graceful fallback path.
• Validated omnichannel cart builder for the matched products.
• 6-stage per-stage provenance (skin_analysis, skin_tone, makeup_vto, apparel_vto, regimen_engine, cart_builder): provider, live/simulated mode, status, units reserved and spent.
• Thread-safe atomic 1,000-unit budget ledger that prevents credit exhaustion and double-charging on non-idempotent POST timeouts.
• Privacy: in-memory EXIF stripping before any upload.

How we built it

Python, the Perfect Corp YouCam API v2.0, CIELAB/CIEDE2000 colour science, and a JSON-RPC 2.0 Model Context Protocol (MCP) server so AI agents can drive the whole concierge. Designed with Claude Opus 5.5, reviewed with OpenAI Codex GPT-5.6 Luna, implemented with Sakura AI Engine.

Challenges

Keeping credit spend safe under timeouts, making colour matching perceptually correct, and ensuring recommendations never conflict with skin-barrier contraindications.

What's next

More live YouCam runs, more product catalogs, and additional try-on categories.

Built With

• ai-agent
• cielab
• ecommerce
• mcp
• perfect-corp
• python
• skin-analysis
• virtual-try-on
• youcam-api