AI-native life rhythm coordination assistant for intimate groups.
一款以 AI 为原生驱动、服务于亲密团体（情侣、家庭、好友）的生活节律协同助手。
Live Demo: https://maichong.rxcloud.group
| Feature | Description |
|---|---|
| Apple/iOS Design | Authentic iOS design system following Human Interface Guidelines |
| Collaborative Timelines | Create shared timelines with your partner, family, or friends |
| AI Chat Assistant | Tell the AI your plans in natural language and it creates events automatically |
| Realtime Sync | Changes are synced instantly across all members via Supabase Realtime |
| Share Cards | Generate beautiful screenshot cards of your schedule to share |
| Invite via Link | Share an invite link to add members to your timeline |
| Demo Mode | Try the app without signing up (data stored in localStorage) |
| Layer | Technology |
|---|---|
| Frontend | Vite + Vanilla JS (ES Modules) |
| Styling | CSS Variables, Apple/iOS design system |
| Icons | Lucide (linear stroke icons, tree-shakeable) |
| Backend | Supabase (Auth + PostgreSQL + Realtime) |
| AI | GLM-4 (OpenAI-compatible API) |
| Screenshots | modern-screenshot |
| Deployment | Vercel |
# Install dependencies
npm install
# Start dev server (port 3000)
npm run dev
# Production build
npm run build
# Run E2E tests (requires Chrome)
node scripts/test-e2e.mjsCopy .env.example to .env and fill in your keys:
cp .env.example .envVITE_SUPABASE_URL=https://your-project.supabase.co
VITE_SUPABASE_ANON_KEY=your-anon-key
VITE_GLM4_API_KEY=your-glm4-api-keyWithout .env, the app runs in demo mode using localStorage.
没有 .env 时，应用会以演示模式运行，数据存储在 localStorage。
src/
├── lib/ # Framework layer (store, router, DOM helpers)
├── services/ # Business logic (auth, timeline, events, AI, realtime)
├── views/ # Page-level views (auth, home, timeline, chat, share, profile)
├── components/ # Reusable UI (header, tab-bar, cards, modals, icons, toast)
├── styles/ # CSS modules (variables, layout, chat, forms, tab-bar...)
├── main.js # App entry point
├── router.js # Hash-based SPA router with guards
└── config.js # Environment config
- No framework — Vanilla JS with a custom reactive store and hyperscript DOM helpers
- Three-layer architecture —
lib/(zero domain knowledge) →services/(business logic) →views/+components/(presentation) - Graceful degradation — Falls back to localStorage when Supabase is not configured
- Apple/iOS Design System — Following Human Interface Guidelines with iOS System Blue (#007AFF), 8pt grid, SF Pro typography scale
Custom pub/sub reactive store with granular key-level subscriptions. Services write state via setState(partial), views subscribe via subscribe(key, cb). Supabase Realtime pushes postgres_changes events over WebSocket for multi-device sync. In demo mode, gracefully degrades to localStorage persistence.
Built following Apple Human Interface Guidelines:
| Token | Value | Description |
|---|---|---|
| Primary Color | #007AFF |
iOS System Blue |
| Success | #34C759 |
iOS Green |
| Warning | #FF9500 |
iOS Orange |
| Danger | #FF3B30 |
iOS Red |
| Tab Bar Height | 49px |
iOS standard |
| Touch Target | 44px |
iOS minimum |
| Spacing | 8pt grid | iOS spacing system |
| Typography | SF Pro scale | Large Title → Caption |
- Imperceptible shadows — Apple uses extremely subtle shadows
- iOS Messages bubbles — Gray for AI, blue for user
- No decorative elements — True minimalism per Apple philosophy
| Hash Route | View | Tab |
|---|---|---|
#/auth |
Login/signup | hidden |
#/ |
Home (timeline list) | Home |
#/timeline/:id |
Timeline events | Timeline |
#/timeline/:id/chat |
AI assistant | AI Chat |
#/profile |
User profile | Profile |
#/timeline/:id/share |
Share card export | hidden |
#/join/:code |
Process invite link | hidden |
Schema: supabase/migrations/001_initial_schema.sql
Tables: profiles, timelines, timeline_members, events, chat_messages — all with Row Level Security policies.
The AI assistant follows a Structured Output + Action Execution pattern (ReAct-lite: single-turn Reason→Act, no iterative tool loop).
How it works:
- Context Assembly — System prompt is dynamically built with timeline title, current events (with IDs), today's date, and a JSON output schema
- LLM Reasoning — GLM-4 (智谱AI, OpenAI-compatible API) receives system prompt + recent 10 conversation messages + user input
- Structured Output — LLM returns
{reply: string, action: object|null}JSON - Action Execution — If
actionis present, the executor dispatches to Event Service (create_event/update_event/delete_event) - Optimistic Update — Store updates immediately, Supabase persists, Realtime syncs to all members
Example flow: User says "周六下午去喝咖啡" → AI Service builds context-aware prompt → GLM-4 reasons about date/time → returns structured JSON with create_event action → Event Service creates the event → Store updates → all members see the new event in realtime.
Fallback paths:
- No API Key → Mock response engine with keyword matching (demo-friendly)
- No Supabase → Chat history persisted to localStorage
# Capture screenshots for docs
node scripts/screenshot.mjs
# Run E2E tests (62 tests)
node scripts/test-e2e.mjsContributions are welcome! Please feel free to submit a Pull Request.
- Fork the repository
- Create your feature branch (
git checkout -b feature/amazing-feature) - Commit your changes (
git commit -m 'feat: add amazing feature') - Push to the branch (
git push origin feature/amazing-feature) - Open a Pull Request