Skip to main content

Getting Started with Realtime Messaging

Pick the path that matches your stack and what you're building.

Build with AI Tools

Use our Realtime Messaging reference file, or the smaller file for your stack (JavaScript SDK, Python SDK, Flutter SDK or raw WebSocket), with Claude Code, Cursor, or any AI coding tool to build your real-time app faster. Copy-paste prompts available on the Build with AI page. Never paste a secret key into an AI tool: keep it in an environment variable on your server and tell the tool the variable's name.

Pick your path​

Using JavaScript or TypeScript?​

Use the @metered-ca/realtime SDK. It handles framing, reconnect, perfect-negotiation WebRTC, and auto-applies the TURN credentials the signalling server injects. Five lines to connect-subscribe-publish, ~30 lines for a working video call.

npm install @metered-ca/realtime

→ JavaScript SDK — Getting Started

Using any other stack?​

Speak the wire protocol directly — Go, Python, Java, Swift, Kotlin, Rust, Unity, anything that can open a WebSocket and sign HS256 JWTs.

→ Introduction → Quickstart — connect, subscribe, and publish in under 30 lines (with Browser and Node tabs)

The SDK and raw-WS clients speak the same protocol byte-for-byte. An SDK client and a raw-WS client can be peers in the same channel.

Just want a copy-paste demo?​

The Single-File HTML Demo runs in one HTML file with no build, no backend, no JWT minting. Demonstrates presence, chat, WebRTC video, and auto-injected TURN in ~200 lines. Save the file, replace the pk_live_ key, open in two browser tabs.

Pick by what you're building​

You're building…Read this first
WebRTC video call with the SDKJS SDK: WebRTC Video Call
WebRTC from a static site (no backend)JS SDK: WebRTC No Backend
WebRTC from a non-JS stackRaw-WS WebRTC Signalling
Chat / classroom / livestream chatJS SDK: Presence & Chat
AI agent communicationJS SDK: AI Agent Communication
IoT telemetryJS SDK: IoT Telemetry
Migrating from simple-peer or PeerJSMigration from simple-peer · from PeerJS

Going to production​

Two pages worth reading before you ship, regardless of which path you took above:

  • Reconnect Best Practices — handling drops, refreshing tokens, surfacing "reconnecting" UI. The single most-skipped doc that bites everyone in production.
  • Limits & Quotas — message size cap, channel cap, per-connection rate limit, plan quotas. Bookmark when sizing a deployment.

If something doesn't work the way you expect, the Troubleshooting guide is symptom-first ("messages aren't arriving", "video stays black", "stuck in reconnecting") rather than code-first.

Reference material​