Real-Time Sync

Implement real-time communication for live updates, collaboration, and presence awareness across applications.

When to Use

Use this skill when building:

• LLM streaming interfaces - Stream tokens progressively (ai-chat integration)
• Live dashboards - Push metrics and updates to clients
• Collaborative editing - Multi-user document/spreadsheet editing with CRDTs
• Chat applications - Real-time messaging with presence
• Multiplayer features - Cursor tracking, live updates, presence awareness
• Offline-first apps - Mobile/PWA with sync-on-reconnect

Protocol Selection Framework

Choose the transport protocol based on communication pattern:

Decision Tree

ONE-WAY (Server → Client only)
├─ LLM streaming, notifications, live feeds
└─ Use SSE (Server-Sent Events)
   ├─ Automatic reconnection (browser-native)
   ├─ Event IDs for resumption
   └─ Simple HTTP implementation

BIDIRECTIONAL (Client ↔ Server)
├─ Chat, games, collaborative editing
└─ Use WebSocket
   ├─ Manual reconnection required
   ├─ Binary + text support
   └─ Lower latency for two-way

COLLABORATIVE EDITING
├─ Multi-user documents/spreadsheets
└─ Use WebSocket + CRDT (Yjs or Automerge)
   ├─ CRDT handles conflict resolution
   ├─ WebSocket for transport
   └─ Offline-first with sync

PEER-TO-PEER MEDIA
├─ Video, screen sharing, voice calls
└─ Use WebRTC
   ├─ WebSocket for signaling
   ├─ Direct P2P connection
   └─ STUN/TURN for NAT traversal

Protocol Comparison

| Protocol | Direction | Reconnection | Complexity | Best For | |----------|-----------|--------------|------------|----------| | SSE | Server → Client | Automatic | Low | Live feeds, LLM streaming | | WebSocket | Bidirectional | Manual | Medium | Chat, games, collaboration | | WebRTC | P2P | Complex | High | Video, screen share, voice |

Implementation Patterns

Pattern 1: LLM Streaming with SSE

Stream LLM tokens progressively to frontend (ai-chat integration).

Python (FastAPI):

python
from sse_starlette.sse import EventSourceResponse

@app.post("/chat/stream")
async def stream_chat(prompt: str):
    async def generate():
        async for chunk in llm_stream:
            yield {"event": "token", "data": chunk.content}
        yield {"event": "done", "data": "[DONE]"}
    return EventSourceResponse(generate())

Frontend:

typescript
const es = new EventSource('/chat/stream')
es.addEventListener('token', (e) => appendToken(e.data))

Reference references/sse.md for full implementations, reconnection, and event ID resumption.

Pattern 2: WebSocket Chat

Bidirectional communication for chat applications.

Python (FastAPI):

python
connections: set[WebSocket] = set()

@app.websocket("/ws")
async def websocket_endpoint(websocket: WebSocket):
    await websocket.accept()
    connections.add(websocket)
    try:
        while True:
            data = await websocket.receive_text()
            for conn in connections:
                await conn.send_text(data)
    except WebSocketDisconnect:
        connections.remove(websocket)

Reference references/websockets.md for multi-language examples, authentication, heartbeats, and scaling.

Pattern 3: Collaborative Editing with CRDTs

Conflict-free multi-user editing using Yjs.

TypeScript (Yjs):

typescript
import * as Y from 'yjs'
import { WebsocketProvider } from 'y-websocket'

const doc = new Y.Doc()
const provider = new WebsocketProvider('ws://localhost:1234', 'doc-id', doc)
const ytext = doc.getText('content')

ytext.observe(event => console.log('Changes:', event.changes))
ytext.insert(0, 'Hello collaborative world!')

Reference references/crdts.md for conflict resolution, Yjs vs Automerge, and advanced patterns.

Pattern 4: Presence Awareness

Track online users, cursor positions, and typing indicators.

Yjs Awareness API:

typescript
const awareness = provider.awareness
awareness.setLocalState({ user: { name: 'Alice' }, cursor: { x: 100, y: 200 } })
awareness.on('change', () => {
  awareness.getStates().forEach((state, clientId) => {
    renderCursor(state.cursor, state.user)
  })
})

Reference references/presence-patterns.md for cursor tracking, typing indicators, and online status.

Pattern 5: Offline Sync (Mobile/PWA)

Queue mutations locally and sync when connection restored.

TypeScript (Yjs + IndexedDB):

typescript
import { IndexeddbPersistence } from 'y-indexeddb'
import { WebsocketProvider } from 'y-websocket'

const doc = new Y.Doc()
const indexeddbProvider = new IndexeddbPersistence('my-doc', doc)
const wsProvider = new WebsocketProvider('wss://api.example.com/sync', 'my-doc', doc)

wsProvider.on('status', (e) => {
  console.log(e.status === 'connected' ? 'Online' : 'Offline')
})

Reference references/offline-sync.md for conflict resolution and sync strategies.

Library Recommendations

Python

WebSocket:

• websockets 13.x - AsyncIO-based, production-ready
• FastAPI WebSocket - Built-in, dependency injection
• Flask-SocketIO - Socket.IO protocol with fallbacks

SSE:

• sse-starlette - FastAPI/Starlette, async, generator-based
• Flask-SSE - Redis backend for pub/sub

Rust

WebSocket:

• tokio-tungstenite 0.23 - Tokio integration, production-ready
• axum WebSocket - Built-in extractors, tower middleware

SSE:

• axum SSE - Native support, async streams

Go

WebSocket:

• gorilla/websocket - Battle-tested, compression support
• nhooyr/websocket - Modern API, context support

SSE:

• net/http (native) - Flusher interface, no dependencies

TypeScript

WebSocket:

• ws - Native WebSocket server, lightweight
• Socket.io 4.x - Auto-reconnect, fallbacks, rooms
• Hono WebSocket - Edge runtime (Cloudflare Workers, Deno)

SSE:

• EventSource (native) - Browser-native, automatic retry
• Node.js http (native) - Server-side, no dependencies

CRDT:

• Yjs - Mature, TypeScript/Rust, rich text editing
• Automerge - Rust/JS, JSON-like data, time-travel

Reconnection Strategies

SSE: Browser's EventSource handles reconnection automatically with exponential backoff. WebSocket: Implement manual exponential backoff with jitter to prevent thundering herd.

Reference references/sse.md and references/websockets.md for complete implementation patterns.

Security Patterns

Authentication: Use cookie-based (same-origin) or token in Sec-WebSocket-Protocol header. Rate Limiting: Implement per-user message throttling with sliding window.

Reference references/websockets.md for authentication and rate limiting implementations.

Scaling with Redis Pub/Sub

For horizontal scaling, use Redis pub/sub to broadcast messages across multiple backend servers.

Reference references/websockets.md for complete Redis scaling implementation.

Frontend Integration

React Hooks Pattern

SSE for LLM Streaming (ai-chat):

typescript
useEffect(() => {
  const es = new EventSource(`/api/chat/stream?prompt=${prompt}`)
  es.addEventListener('token', (e) => setContent(prev => prev + e.data))
  return () => es.close()
}, [prompt])

WebSocket for Live Metrics (dashboards):

typescript
useEffect(() => {
  const ws = new WebSocket('ws://localhost:8000/metrics')
  ws.onmessage = (e) => setMetrics(JSON.parse(e.data))
  return () => ws.close()
}, [])

Yjs for Collaborative Tables:

typescript
useEffect(() => {
  const doc = new Y.Doc()
  const provider = new WebsocketProvider('ws://localhost:1234', docId, doc)
  const yarray = doc.getArray('rows')
  yarray.observe(() => setRows(yarray.toArray()))
  return () => provider.destroy()
}, [docId])

Reference Documentation

For detailed implementation patterns, consult:

• references/sse.md - SSE protocol, reconnection, event IDs
• references/websockets.md - WebSocket auth, heartbeats, scaling
• references/crdts.md - Yjs vs Automerge, conflict resolution
• references/presence-patterns.md - Cursor tracking, typing indicators
• references/offline-sync.md - Mobile patterns, conflict strategies

Example Projects

Working implementations available in:

• examples/llm-streaming-sse/ - FastAPI SSE for LLM streaming (RUNNABLE)
• examples/chat-websocket/ - Python FastAPI + TypeScript chat
• examples/collaborative-yjs/ - Yjs collaborative editor

Testing Tools

Use scripts to validate implementations:

• scripts/test_websocket_connection.py - WebSocket connection testing