# Signal Native — app-styled homepage comparison

## Intent and reference anchors

Keep the Confidence Lab story, content and section sequence so Joshua can judge the styling directly. Publish independently at `/redesigns/confidence-app/`; preserve `/redesigns/confidence-lab/` and `/` byte-for-byte. Direction: **Signal Native**, the app's restrained technical product UI adapted to a marketing page. Anchors: StudySignal's test picker and quiz, its shared card/button/signal components, and its Space Grotesk + Geist + Geist Mono type system. These actual product references take precedence over generic skill advice against familiar fonts, gradients or glow.

Motion **3/10** (control feedback, no reveal choreography), density **3/10** (the existing story with more compact typography), polish **8/10** (precise app surfaces and two complete themes).

## Tokens — before layout and implementation

Import `/app.css` followed by `/signal-theme.css`. They own all `--stl-*` colors, typography, radii and motion. Do not redefine that system. Reuse `stl-card`, `stl-card--hero`, `stl-choice`, `stl-choice__letter`, `stl-signal__*`, and `stl-btn--pill-rainbow`. Candidate CSS owns layout only and adaptations specific to marketing illustrations.

Dark canvas `#050509`, wells `#0a0b12`, elevated interiors `#0d0e18`; text `#f0f2f7`, muted `#a8b0c4`, metadata `#787e92`. Light canvas `#f3f4f8`, wells `#fafbfd`, text `#16181f`, muted `#3f4555`, metadata `#5e6573`. Signal marks use the app's green/yellow/red and RGB tokens. Signal text uses its contrast-adjusted `-text` tokens. Cool ambient petrol and blue match the app's resting background, fading out within the hero.

| Pair           | Foreground | Background | WCAG    | Foreground OKLCH           |
| -------------- | ---------- | ---------- | ------- | -------------------------- |
| dark body      | `#f0f2f7`  | `#0d0e18`  | 17.15:1 | `oklch(0.961 0.007 268.5)` |
| dark muted     | `#a8b0c4`  | `#0d0e18`  | 8.85:1  | `oklch(0.757 0.030 268.3)` |
| dark metadata  | `#787e92`  | `#0d0e18`  | 4.76:1  | `oklch(0.595 0.031 272.2)` |
| dark mastery   | `#86efac`  | `#0d0e18`  | 13.68:1 | `oklch(0.871 0.136 154.4)` |
| light body     | `#16181f`  | `#f3f4f8`  | 16.13:1 | `oklch(0.210 0.014 272.6)` |
| light muted    | `#3f4555`  | `#f3f4f8`  | 8.71:1  | `oklch(0.391 0.028 268.9)` |
| light metadata | `#5e6573`  | `#f3f4f8`  | 5.33:1  | `oklch(0.506 0.024 264.3)` |
| light mastery  | `#15803d`  | `#f3f4f8`  | 4.56:1  | `oklch(0.527 0.137 150.1)` |

Typography: self-hosted **Space Grotesk 500/600/700** for headings; **Geist 400/500/600/700** for body; **Geist Mono 400/500** for metadata and numbers. Base scale uses ~1.25: 12/14 optical metadata, 16 body, 20, 24, 32, 40, 48, 64. Hero 64 desktop / 48 tablet / 40 phone; body 16, lede 18. Metadata down to 11 only for nonessential question-index labels. App wordmark paths retained with currentColor for themes.

Spacing: 4/8/12/16/20/24/32/40 plus marketing section gaps 48/64/80/96. Shell 1120 desktop, gutters 32/24/20. Radii reuse 8/10/14/18/pill. Thin borders keep their width in every state. Motion: 120/200/280ms from app tokens; reduced-motion removes animation and smooth scrolling.

## IA and layout skeleton

1. App-like 64px sticky header: wordmark, in-page links, light/dark control, sign-in and outlined practice CTA.
2. Two-column hero: smaller Space Grotesk headline; a quieter score/signal panel with product-style tinted cells and segmented pills.
3. Exam availability strip using compact rounded chips.
4. Method explanation; three semantic signal states, presented as one connected strip.
5. Same working question demo, inside the actual app's card and answer/signal controls. Theme-consistent surface replaces the cream inversion.
6. Progress and learning loop; app-style progress panel and numbered rows.
7. Teacher comparison; neutral panel, confidence marks and same example students.
8. Founder, FAQ, closing CTA and footer, with the same content as the original.

## Component behavior

Native radio group updates the shared choice selected class, then one of three confidence buttons reveals all six original outcomes. Reset clears state and restores focus. Native disclosures, menu Escape, visible focus, and 44px minimum touch targets. Theme toggle starts in dark, keeps its preference session-local to this comparison, and does not change the production app's stored theme. No new backend calls, accounts, analytics or payment behavior.

## Direction choice and critique plan

A simple palette swap would keep the oversized editorial letterspacing and cream band, failing brand consistency. A dashboard shell would constrain the story. The selected option uses the full app type/control/surface system while retaining marketing hierarchy. A→B and B→A review favor it on consistency and clarity; preserving content makes comparison meaningful.

Three rejection risks: (1) duplicated style tokens drift from the app — import the real stylesheets; (2) green labels lose contrast in light mode — use `-text` tokens and test both themes; (3) polished surfaces become nested boxes — only the actual demos receive card chrome, keep prose open. Grayscale hierarchy must hold through scale, weight and spacing. Remove ornament: no hero tab decoration, no entry choreography, no unrelated imagery. Exceptions to generic design-skill rules are intentional product matching: the existing signal outline gradient, native signal-dot glow, and the existing three-font functional hierarchy.
