Run your first agent - Agentium

You need Node.js 22.18+ within v22 or 24.11+ within v24 and an OpenAI API key. This example targets Agentium 4.0.0. Follow installation to add matching packages and optional integrations.

Set up a project

mkdir agentium-support
cd agentium-support
npm init -y
npm pkg set type=module
npm install @agentium/core@4.0.0 openai zod
npm install --save-dev typescript tsx @types/node
export OPENAI_API_KEY="your-key"
export OPENAI_MODEL="gpt-6.1-sol"

Choose a model available to your account. Keep API keys in your server environment. For other providers, see the integration directory.

Create the Agent

Save this as support.ts:

support.ts

import { Agent, openai } from "@agentium/core";

const agent = new Agent({
  name: "support",
  model: openai(process.env.OPENAI_MODEL ?? "gpt-6.1-sol"),
  instructions:
    "Help customers with orders. Ask for an order ID before discussing an order.",
});

try {
  const result = await agent.run("Where is my order?");
  if (result.status !== "completed") {
    throw new Error(`Run ended with status: ${result.status}`);
  }
  console.log(result.text);
} finally {
  await agent.close();
}

Run it

npx tsx support.ts

You should see a request for your order ID. The wording varies. This Agent has no order database yet; the next step is to give it a lookup tool. If the request fails, check the API key, model access, package build, and terminal status. The troubleshooting guide explains how to isolate the failing boundary.

Add your first capability

Look up an order

Continue with a validated tool and a verifiable result.

Understand the execution

See how model calls and tool results form a run.