Ship your first service.
One command. Real HTTPS. The same platform you'd use in production.
$ npx @computesphere/csph deployConnect your AI assistant
Add ComputeSphere to Claude or Cursor and ask it to deploy your service. The assistant handles auth, picks the right commands, and reports your live URL — no terminal required.
Install csph
Pick one. csph uses device-code OAuth by default — works inside AI agents, CI, and SSH sessions with no browser required.
$ brew install computesphere/cli/csphThen authenticate. The default flow prints a code and a URL — paste the code in your browser to confirm.
$ csph auth loginDeploy
Pick any container image — public or private. ComputeSphere schedules it, terminates TLS, and exposes it on a *.computesphere.app URL.
$ csph deploy --image ghcr.io/computesphere/hello:latestWhen it's done, you'll see your service's live URL — something like:
https://silver-cobra.computesphere.appcurl it. Open it. Paste it into Slack.
You're live.
HTTPS
Auto-issued and renewed by the platform. No cert wrangling.
Zero-downtime
Rolling deploys with health checks. Old spherelet stays until the new one passes.
Logs
Stream stdout/stderr live in the console or via csph logs.
Metrics
Request rate, latency, error rate, spherelet CPU and memory — out of the box.
Next
- Add a database, cache or queue: SphereDB (PostgreSQL), SphereCache (Valkey) or SphereQueue (RabbitMQ)
- Custom domain — bring your own hostname with auto-renewed TLS
- Concepts — account, project, environment, service, spherelet, deployment