SCS AWS Kiro Hackathon 2026

This is the application designed for the hackathon hosted by Singapore Computer Society with AWS, showcasing Kiro.

No license.

Mandai WildDex & Ranger Expeditions

A demo-mode, offline-capable iOS PWA. Four tabs — Map, Camera, Quests, Journal — running entirely locally against LocalStorage and mock services. No backend, no auth, and no real AWS calls at any point. The "AI" scan is a local DemoAIAnalyzer behind an AIAnalyzer interface (see src/services/aiDemo.ts); a future AWS-backed implementation would satisfy the same interface but is intentionally not built here.

Requirements

Node 20.x (see .nvmrc) — nvm use
npm

Setup

npm install
npm run dev      # dev server
npm run build    # tsc --noEmit + vite build (emits dist/ + service worker)
npm run preview  # serve the production build (needed to exercise the service worker)

PWA & offline

Service worker generated by vite-plugin-pwa (generateSW); precaches the app shell, bundled JS/CSS, and static assets (icons, splash, species/map images).
Because the app makes no network calls, precaching the shell + bundled data is sufficient for full offline function — no runtime caching rules are needed.
iOS meta tags + manifest.json configure standalone display, theme color, and the install experience. Icons/splash use the supplied art in public/icons and public/splash.
First-run overlay explains iOS "Add to Home Screen"; dismissal is persisted (hasSeenInstallPrompt).

Accessibility

Reduced motion honored via a single useReducedMotion() hook (combines the in-app setting and the OS prefers-reduced-motion query); consumed by every animation.
ARIA roles/labels on toasts, progress bars, badges, and interactive controls; 44×44px minimum touch targets.
Sound and reduced-motion are user-toggleable from the settings menu.

Verification status (per Requirement 9.4)

Verified in this environment:

npm run build and tsc --noEmit complete with zero errors.
localStorage access is confined to src/services/storage.ts (no component touches it directly).
No raw camera-frame/base64 data is persisted to state — only species IDs.
Rank thresholds and XP calibration reviewed against the fixed table (a full playthrough crosses into Master Ranger).

Configured, untested (requires a real browser/device — not verifiable headlessly):