Stateful, Scalable, Realtime Backend Framework

Intro

The modern way to build multiplayer, realtime, or AI agent backends.

Supports Rivet, Cloudflare Workers, Bun, and Node.js.

Architecture

• 💾 Durable, In-Memory State: Fast in-memory access with built-in durability — no external databases or caches needed.
• ⚡ Ultra-Fast State Updates: Real-time state updates with ultra-low latency, powered by co-locating compute and data.
• 🔋 Batteries Included: Integrated support for state, RPC, events, scheduling, and multiplayer — no extra boilerplate code needed.
• 🖥️ Serverless & Scalable: Effortless scaling, scale-to-zero, and easy deployments on any serverless runtime.

Features

• 💾 State: Fast in-memory access with built-in durability.
• 💻 RPC: Remote procedure calls for seamless client-server communication.
• 📡 Events: Real-time event handling and broadcasting.
• ⏰ Scheduling: Timed tasks and operations management.
• 🌐 Connections & Multiplayer: Manage connections and multiplayer interactions.
• 🏷️ Metadata: Store and manage additional data attributes.

What makes ActorCore different?

ActorCore is the modern way to build realtime, stateful backends.

Feature	ActorCore	Durable Objects	AWS Lambda	Redis	Socket.io
In-Memory State	✓	✓		✓	✓
Durable State	✓	✓
RPC	✓	✓	✓		✓
Events	✓				✓
Scheduling	✓
Edge Computing	✓ †	✓	✓
No Vendor Lock	✓			✓	✓

† = on supported platforms

Getting Started

Step 1: Installation

# npm
npm add actor-core

# pnpm
pnpm add actor-core

# Yarn
yarn add actor-core

# Bun
bun add actor-core

Step 2: Create an Actor

import { Actor, type Rpc } from "actor-core";

export interface State {
  messages: { username: string; message: string }[];
}

export default class ChatRoom extends Actor<State> {
  // initialize this._state
  _onInitialize() {
    return { messages: [] };
  }

  // receive an remote procedure call from the client
  sendMessage(rpc: Rpc<ChatRoom>, username: string, message: string) {
    // save message to persistent storage
    this._state.messages.push({ username, message });

    // broadcast message to all clients
    this._broadcast("newMessage", username, message);
  }
}

Step 3: Connect to Actor

import { Client } from "actor-core/client";
import type ChatRoom from "../src/chat-room.ts";

const client = new Client(/* manager endpoint */);

// connect to chat room
const chatRoom = await client.get<ChatRoom>({ name: "chat" });

// listen for new messages
chatRoom.on("newMessage", (username: string, message: string) =>
  console.log(`Message from ${username}: ${message}`)
);

// send message to room
await chatRoom.sendMessage("william", "All the world's a stage.");

Step 4: Deploy

Deploy to your platform of choice:

• Rivet
• Cloudflare Workers
• Bun
• Node.js

Community & Support

• Join our Discord
• Follow us on X
• Follow us on Bluesky
• File bug reports in GitHub Issues
• Post questions & ideas in GitHub Discussions

License

Apache 2.0