# 08 · Authenticate

**Screen:** `data-screen="authenticate"` · child mode

## Intent
Brand-defense as a sensory tool. A piece comes in for return or repair — scan it, watch the pipeline run, get a verdict. Sells "brand defense at every layer" as something tactile.

## Key elements
- Viewfinder (4:5) with scanning crosshair corners and a moving cream scanline
- 4-step pipeline list: stitch density · hardware · leather grain · cert ID lookup
- Demo-only toggle above **Begin scan**: "Simulate a counterfeit result"
- **Begin scan** primary CTA runs the pipeline (each step ~140–480ms)
- Genuine path: scanline hides, "Authentic · matched" result panel reveals — "Speedy 20 · 2019 · paired to Mme. Laurent's passport"
- Counterfeit path (toggle on): step 1 (stitch density) shows `MISMATCH` in the alert color instead of a normal timing readout; result panel switches to the alert palette — "Flagged · pattern mismatch," routed to brand-defense review, explicitly not touching Mme. Laurent's own passport

## Notes
Built 2026-07-12: previously the scan could only ever resolve positive, so the app never actually demonstrated the "brand defense" claim it makes everywhere else — this closes that gap. The counterfeit tell (stitch density) deliberately echoes the brand-defense feed's own "stitch density mismatch" copy on the Tokyo listing, so the two screens read as the same detection engine, not two unrelated features.

Deep-link note: the walkthrough shows both the genuine and counterfeit results as separate phones, but they're the same `data-screen="authenticate"` element — a plain `#authenticate` hash always lands on the untouched pre-scan default. The counterfeit phone links to `#authenticate-counterfeit` instead, a routing alias (see `HASH_ALIASES` in the deep-link routing IIFE) that snaps to the real `authenticate` screen and then runs the pipeline with the toggle pre-set, so the link actually shows what it screenshots. Same fix shape as Intel's `askResult-q3`/`-renewal`/`-fab` aliases — caught during the 2026-07-12 prototype review cycle, not before.

## Real-world precedent
The closest real analogs are The RealReal's TRR Vision/TRR Shield (ML triage on 50+ construction/material metrics before a human authenticator looks at the item) and Entrupy's portable scanner (200+ macro/micro images, under 5 minutes, ~20 brands in its reference database). This screen's 4-step pipeline is a compressed, stylized version of that same "machine screens first, verdict is explainable" workflow. See `LV/Real-World-Inspiration-Research.md` §5.