Docs / Quickstart

Deploy your first agent.

By the end of this page you'll have an always-on agent with its own cloud machine. You'll need a HyperCLI account and Node 18+.

About 5 minutes

Install the CLI~30 sec

$npm install -g hypercli

Opens your browser, links the CLI to your account. Your trial starts here — card down, agent up.

$hyper configure

✓ Signed in

Create your agent~90 sec

A name, a model, and what your agent should care about — that becomes its standing instructions. Change everything later.

$hyper agents create my-agent

? Model: kimi-k2.6

? Purpose: Watch my repo, summarize activity every morning

✓ Created my-agent

Deploy~2 min

Provisions your agent's machine, wires up inference, connects channels. This is the config work other platforms make you do — done for you.

$hyper agents start my-agent

Provisioning machine........... done

Wiring inference (K2.6 + Qwen)... done

→ my-agent is live

Say hello~30 sec

$hyper agents shell my-agent

my-agent: Hey. I'm watching the repo — first summary lands at 7am. Need anything before then?

That's it. It stays on when you close the terminal — that's the point.

You now have an always-on machine with browser, voice, media generation, and memory — every capability, no add-ons. It's already remembering this conversation.

Where to next.

Put it on a schedule

Your agent can write its own cron jobs — just ask it, or set one explicitly.

Give it a voice

Clone yours, design one from a description, or pick a stock voice.

Add channels

Reach it from Slack, Teams, Telegram, or WhatsApp.

Just want inference?

Skip the agent — grab an API key and point your OpenAI or Anthropic code at us.

Stuck? Ask your agent.

It can read these docs. The full reference lives at docs.hypercli.com — the source of truth this page defers to.

Read the full quickstart