> ## Documentation Index
> Fetch the complete documentation index at: https://internal.september.wtf/llms.txt
> Use this file to discover all available pages before exploring further.

# Quickstart

> Make your first request to the Engine — in curl, Python, or Node.

This page sends your first request to the Engine. By the end, you'll have a
streaming response from a real agent, in the language of your choice.

If you don't yet have an Engine running, follow [Local development
quickstart](../../operations/local-dev/quickstart) first.

## Pick your language

<CardGroup cols={3}>
  <Card title="curl" icon="terminal" href="./quickstart-curl">
    No dependencies, just a shell.
  </Card>

  <Card title="Python" icon="python" href="./quickstart-python">
    The most common path for AI development.
  </Card>

  <Card title="Node / TypeScript" icon="node" href="./quickstart-node">
    For web and agent backends.
  </Card>
</CardGroup>

## What you'll need

* The URL of an Engine you can reach. For local development that's
  `http://localhost:8000`.
* An Engine API key. Set as `X-Engine-Key` on every request.
* One open task slot. The Engine accepts an arbitrary `task_id` you pick.

## The shape of the call

Every quickstart in this section makes the same request:

```http theme={null}
POST /execute
X-Engine-Key: <your-key>
Content-Type: application/json

{
  "message": "Say hello in one short sentence.",
  "task_id": "demo-001"
}
```

And consumes the same Server-Sent Event stream that comes back. The only
difference is which language and HTTP client you're using.

When you're done, head to [the agent loop](../agents-and-tools/agent-loop)
to understand what happens between your request and the response, or
straight to [the API reference](../api-reference/overview) if you'd rather
read the contract first.