Claude Code Goes Mobile with Remote Control
I've been running Claude Code on a headless Intel NUC in my office and connecting from my phone via SSH through Tailscale for months. So when Anthropic shipped Remote Control this week, my first reaction wasn't amazement. It was recognition.
Welcome to the party, everyone else.
What Remote Control Actually Does
Remote Control is a synchronization layer that bridges a local Claude Code CLI session with the Claude mobile app or web interface at claude.ai/code. You start a task in your terminal, it stays running on your machine, and you pick it up from your phone on the couch. If your laptop sleeps or your network drops, the session reconnects automatically when the machine comes back online.
That last part is the key detail. This isn't screen sharing or a remote desktop hack. Claude keeps running locally, with full access to your filesystem, MCP servers, environment variables, and project config. The phone is just a window into the session, not a separate environment. All traffic routes through the Anthropic API over TLS, with short-lived credentials scoped to a single purpose. Your machine never opens an inbound port.
The docs describe it cleanly: "Start a task at your desk, then pick it up from your phone on the couch or a browser on another computer."
Setting It Up
The setup is dead simple. Update to the latest version of Claude Code, then run:
claude remote-control
That's it. The terminal displays a session URL and a QR code (press spacebar to toggle it). Scan the QR code with the Claude iOS or Android app, or open the URL in any browser. You can also type /remote-control or /rc inside an existing session to hand it off without losing your conversation history.
There's a useful tip in the docs: use /rename before running /remote-control so the session has a descriptive name in the session list. Small thing, but it matters when you have multiple projects running.
One important caveat: if you close the terminal or kill the claude process, the session ends. This isn't a daemon. It's a foreground process. Run it again to restart. And if your machine stays awake but can't reach the network for more than roughly 10 minutes, the session times out. Those are both real limitations for long-running tasks.
Limitations and Who Should Actually Pay Attention
Remote Control launched in research preview on Pro and Max plans, though availability may still be rolling out. It's explicitly unavailable on Team or Enterprise plans. That's a notable gap if you're part of a team that assumed it would cover the whole organization.
The pricing context matters here too. Max runs $100-200 per month. That's not a casual spend. If you're already on Max for heavy Claude Code usage, Remote Control comes at no extra cost and is worth trying today. If you're on Pro at $20/month, check whether it's reached your account yet; the rollout is still expanding.
Each Claude Code session supports one remote connection at a time. If you run multiple instances, each gets its own session.
The DIY Setup Is Still Better in Some Ways
Here's the honest take: if you're already running Claude Code on a remote machine with SSH and a VPN like Tailscale, the DIY approach gives you things Remote Control doesn't.
You can run multiple Claude Code sessions from the same terminal multiplexer, connect with any SSH client you already trust, use tmux to keep sessions alive even if the SSH connection drops, and do all of this without the Anthropic API as a relay in the middle. My NUC setup costs nothing beyond the hardware. There's no dependency on Anthropic's relay infrastructure staying up, and no plan tier required.
Remote Control's advantage is that it doesn't require any of that setup. No VPN to configure, no server to maintain, no SSH keys to manage. For someone who hasn't built a remote dev environment before, it removes a real chunk of friction. The UX is also better: conversation history syncs across surfaces, you can send messages from terminal or phone interchangeably, and the QR code setup is fast.
The real story isn't the feature itself. It's what the feature represents. Anthropic is making it normal to run untethered AI agents, to kick off long tasks and walk away, to interact with your agent from a different physical context than where it's working. Treating the agent as something that runs in the background and checks in with you, rather than something you sit and watch, is a real shift in how this tooling works.
That shift was already happening for builders who were wiring these things together themselves. Now it's a checkbox in the settings.
If you're on Pro or Max, run claude remote-control tonight and see how it feels. If you've already built your own SSH tunnel setup, you don't need to switch. But notice that what felt like a power-user trick six months ago just shipped as a first-party feature. That's a reliable signal about where everything is heading.
