The CLI
It's in the name.
One binary drives the whole platform — agents, GPUs, media, voice, memory, billing. If HyperCLI can do it, hyper can do it from your terminal.
Everything is a command.
The whole surface, one verb away. Every block links to its reference.
Agents
$hyper agents create my-agent --type openclaw-pro$hyper agents shell my-agenthyper agents reference →Media flows
$hyper flow text-to-image "a product hero image"$hyper flow text-to-video "30s launch teaser"hyper flow reference →GPU jobs
$hyper instances launch train:latest -g h100 -n 2$hyper jobs logs --followinstances + jobs →Plus a full TUI: hyper tui — live jobs, logs, and metrics without leaving the terminal.
Your agent speaks it too.
The CLI isn't just your interface — it's the shared language. The same commands you run are how your agent manages itself.
You, from your laptop
Your agent, from its machine
Anything you can script, it can script. Anything it does, you can read in plain command history — auditable by design, automatable by default.
Scriptable by default.
Everything returns clean output for pipes and scripts — your deploy process is a bash file, your cron jobs are one-liners.
One install, no ceremony.
pip or npm, one auth command, and every capability is live. No plugin maze, no per-service credentials.
SDKs when you outgrow the shell.
Python and TypeScript SDKs mirror the CLI one-to-one — prototype in the terminal, productionize without relearning anything.
Documented like we mean it.
Per-command reference, guides, and OpenAPI specs at docs.hypercli.com — the source of truth this page happily defers to.
Named after the thing
we care about most.
Install it, sign in, and the whole platform is a tab-complete away.
Commands shown are illustrative. The reference at docs.hypercli.com is the source of truth — this page happily defers to it.