Skin AI × the products you already own
Your shelf, checked against your skin.
Before you buy another product: one selfie, the bottles you already own, and an honest answer. Most of the time it's buy less.
or start with a sample
Samples are AI-generated faces, not real people; their scores are real YouCam results. Your photo goes once to YouCam's Skin Analysis API and is not stored by SkinShelf; your shelf stays in this browser. Cosmetic guidance, not medical advice.
1
Your skin
YouCam AI Skin Analysis · 7 concerns–overall
–skin age
Levels: priority under 60 · watch under 80 · fine.
2
Your shelf
ingredient lists, as printed on the bottleAdd a bottle
·
3
The audit
every flag names the ingredient and the score behind itAlready covered by what you own
4
Your week
only what you own · actives never share a night
How it works
- Analyze. Your selfie goes to a small server function that holds the YouCam API key. It uploads the photo with YouCam's File API, runs a
skin-analysistask for seven concerns (wrinkle, pore, texture, acne, redness, moisture, oiliness) and returns the scores and masks. - Read your shelf. Each ingredient list is read exactly as printed against a curated list of 111 INCI names, plus patterns for ceramides and plant essential oils (acids, retinoids, vitamin C, niacinamide, humectants, barrier lipids, UV filters, fragrance, drying alcohols…). No guessing: a name it doesn't know stays "unknown".
- Audit. Plain rules, written down: how many exfoliant and retinoid nights your redness and hydration can take; which pairs shouldn't share a night; which products do the same job; which irritants matter for your scores; what your scores need that nothing on your shelf does.
- Plan. A seven-night week that uses only what you own, with actives spaced out. The verdict is "buy nothing" whenever that's true.
Built by David J for the YouCam API Skin AI & eCommerce VTO Hackathon (Oct 2026), with help from an AI assistant. No AI runs at audit time: the rules are in src/audit.js. Open source (MIT) on GitLab.