Get Started with AI Agents: Hello World

Your first AI agent in under 5 minutes. Two files, one command — no Python, no LangChain, no boilerplate. Just a working agent that runs and returns a result.

What You Will Build

A simple AI agent that greets you and tells you today's date. It sounds trivial, but this is the exact foundation every production agent is built on — a name, instructions, a step, and a config. Once this runs, connecting it to a database, an API, or a file system is just adding more steps.

What You Need

• Node.js 18+ installed (for the npx approach), or the OE Runtime binary downloaded from openenthrium.com/runtime
• An API key from OpenAI, Anthropic, or any other supported LLM provider
• A text editor — nothing else

Create the Skill Files

Create a folder called hello-world/ anywhere on your machine and add two files inside it:

---
name: hello-world
description: Greets you and tells you today's date. Use to verify OE Runtime is working correctly or as a starting template.
license: Apache-2.0
metadata:
  author: Open Enthrium
  version: "1.0"
---

You are a helpful assistant.

## Step 1: Greet
Say hello and today's date.

name: Hello World
description: Greets you and tells you today's date
skills:
  - path: ./
    trigger_type: auto

The SKILL.md holds the personality and steps. The agent.yaml wires the skill to OE Runtime. Both files live in the same hello-world/ folder alongside oe-config.json.

Create oe-config.json

In the same folder, create oe-config.json. This tells the runtime which LLM to use:

{
  "llm": {
    "provider": "openai",
    "model": "gpt-4o",
    "apiKey": "sk-..."
  },
  "server": {
    "enabled": false,
    "port": 3333,
    "apiKey": "your-secret-api-key"
  }
}

Replace sk-... with your actual OpenAI API key. If you prefer Anthropic, set "provider": "anthropic" and "model": "claude-sonnet-5" with your Anthropic key. The agent itself never changes — only the config does.

Run It

Open a terminal in the parent folder (one level above hello-world/) and run:

npx -y @openenthrium/oe-runtime@latest ./hello-world

The first run downloads the OE Runtime binary automatically. Subsequent runs are instant. If you already downloaded the binary, use it directly:

# Windows
oe-runtime-win.exe ./hello-world

# macOS / Linux
./oe-runtime-linux ./hello-world

Within a few seconds you will see output like this:

What Just Happened

The runtime read agent.yaml, loaded your LLM config from oe-config.json, and executed the single step — sending the instruction "Say hello and today's date" to GPT-4o. The LLM responded, the runtime printed the result, and exited cleanly.

No server to start. No SDK to install. No framework to learn. The YAML is the agent.

What's Next

The hello world agent uses only an LLM. Real agents connect to real data. Here are the next things to try:

• Query a database — add a connector to your oe-config.json and ask the agent to run a SQL query. See the SQL database agent guide.
• Call an API — connect to any REST endpoint and have the agent summarise or transform the response. See the REST API agent guide.
• Multi-step workflows — add more steps to agent.yaml to chain actions: fetch data, process it, send a result somewhere.
• Run it on a schedule — use the --cron flag to run the agent automatically at a fixed time.