Managed VMs running Claude Code with Remote Control. Close your laptop. Your sessions stay live on claude.ai/code and your phone. Jean-Claude never sleeps.
logs you in · starts your 3-day free trial · no card · no kicks to the face
claude remote-control is the link between your machine and claude.ai. While the process runs, your sessions sit in the list on every Claude app. When it stops, they are gone. And there is no waking it from the app: the process polls outbound, so claude.ai cannot reach a machine that is off.
On a laptop, that process loses every fight.
An always-on VM can do more than wait for you. Ask Claude to set up a recurring job and it writes the automation for you, or drop in your own markdown file. Each one becomes a long-lived Claude Code session that wakes itself on schedule, does the work, and goes back to idle, sitting right alongside your other sessions in the Claude apps. No laptop open, no process to babysit.
~/.config/jc/automations. The frontmatter sets the cron and the model, the body is the prompt.jclaude automations start reads the folder live and runs one session per file. Sign in once and the VM even starts it for you.Sweep new issues on the repo, label each one, draft a reply, and open a PR when the fix is a one-liner.
Watch CI. The moment a build goes red, pull the logs, find the cause, and push a draft fix before you notice.
Read the overnight logs and leave a tidy summary in the session, waiting for you with your first coffee.
intervals, @macros, or full five-field cron. one long-lived session per file, idle until its cron and restarted if it trips, and because the VM never sleeps the schedule actually holds.
A Jean-Claude VM isn't a bare box you have to housekeep. It's a serviced apartment for your agent: everything pre-installed, everything maintained, concierge included.
Health checks, automatic restarts, image updates, disk headroom: handled. If the claude remote-control process exits, the supervisor has it back before your next prompt. You will never babysit a server at 3am again.
jclaude up provisions a VM in about a minute, Claude Code preinstalled. Sign in once over SSH and the VM registers on claude.ai/code. From then on, it is always on and waiting for you.
Claude Code, codex and opencode preinstalled. Node 24 and the latest Python, with uv handling version switches. Persistent disk, one-command snapshots and rollback, tmux and mosh for the road.
Every VM is isolated, with encrypted disks and scoped tokens. Nothing shared, nothing pooled. Your code keeps its own room.
Closed by default, except 443 straight to :3000. jclaude port open 8080 puts any port on a public URL in about 15 seconds, up to eight at once. The dev server your agent starts is reachable from any device, no tunnels.
Eight regions, so your VM lives next door and keystrokes land as you type them. Mosh and tmux ride along for rough Wi-Fi.
No YAML. No console safari. The whole onboarding is three moves (the tuto counts them for you).
"My agent ran for eleven days straight. My laptop was closed. I was skiing."
"I asked for a VM. I got a VM, plus the distinct feeling that someone in Brussels is personally watching over my cron jobs."
"Honestly, I came for the pun. I stayed for the snapshots."
Start with a free trial. Upgrade when your agent starts asking for more protein.
Free to spar. Not free forever.
One always-on VM. For agents with a day job. And a night job.
On demand. Up to 4 VMs as the mission needs.
Give it a real gym. An always-on VM running Claude Code with Remote Control, a concierge with a beret, and sessions that follow you to your pocket.
free for 3 days. no card required. no roundhouse required.