Quickstart

Install the v2 beta, register a connector, and run an agent.

Install

Requires Node.js 22.13+ or 24+. ESM, CommonJS, and TypeScript declarations are included.

bash

npm install oneringai

Published version: oneringai@2.0.0-beta.1. Both latest and next select this beta. Pin the version for reproducible installs.

Run an agent

Set OPENAI_API_KEY in your environment. Load .env in your application if needed; OneRingAI does not load it automatically. Calls use your provider account.

typescript

import { Agent, Connector, Vendor } from "oneringai";

Connector.create({
  name: "openai-main",
  vendor: Vendor.OpenAI,
  auth: { type: "api_key", apiKey: process.env.OPENAI_API_KEY! },
});

const agent = Agent.create({
  connector: "openai-main",
  model: "gpt-5.6-terra",
  instructions: "Be accurate and concise.",
});

try {
  const response = await agent.run("Explain connector-first architecture.");
  console.log(response.output_text);
} finally {
  agent.destroy();
}

Run with your TypeScript toolchain. To switch providers, register another connector and change connector and model. Supported capabilities vary by model and endpoint.

Add capabilities

Connectors and tools: authenticated external APIs.
Memory: entities, facts, relationships, and retrieval.
V2 controls: evidence, task state, model budgets, and recovery.
Agent runtime: switch complete harnesses. External drivers are preview.
AMOS cookbook: configure a host and test all five drivers.
Examples: offline demonstrations and opt-in live runs.

Upgrade from v1

Replace @everworker/oneringai with oneringai and update imports. Review optional SDK dependencies and enabled features. Existing Agent.create() defaults remain supported where documented.

Migration guide · V2 release notes