You no longer have to choose between going for a walk and having your agent keep working.
Anthropic launched today — February 25, 2026 — one of the most anticipated features by the Claude Code community: Remote Control. And the premise is as simple as it is powerful: you start a coding session in your terminal and control it from your phone, tablet, or any browser. Without losing context. Without restarting. Without being stuck at your desk.
Claude Code’s product manager, Noah Zweben, described it with a phrase that will resonate with many: “Go for a walk, get some sun, take the dog out — without losing the flow.”
How exactly does it work?
The important thing to understand is that Remote Control is not cloud computing. Your code never leaves your machine. What it does is create a secure tunnel (only outgoing HTTPS connections, no open ports) between your local terminal and the Claude app on your phone or browser.
Think of it as a remote desktop for your terminal, but built specifically for AI work.
The setup is minimal:
# Navigate to your project
cd my-project
# Activate Remote Control
claude remote-control
# Or if you're already in an active session
/rc
Your terminal shows a session URL and a QR code. You scan it with your phone and you’re done — you’re seeing and controlling the exact same session that has access to your file system, your MCP servers, and all your local configuration.
If your laptop goes to sleep or the connection drops, the session automatically reconnects when connectivity returns. (Note: if the session drops for more than ~10 minutes, you’ll need to run the command again.)
Real case #1: The 11 PM deploy
The situation: Martina is a backend developer at a startup in Buenos Aires. It’s 10:50 PM, she just launched a deploy and Claude Code is running the test suite + checking for possible regressions in three critical modules. The process will take about 40 minutes.
Before Remote Control: Martina had two options — stay glued to the monitor waiting, or go to sleep and pray nothing breaks.
With Remote Control: Martina activates /rc, scans the QR from her phone, and goes to bed. From the Claude app, she can see in real time what the agent is doing. When Claude finishes a module and needs confirmation to proceed with a delicate change, she gets a notification. Martina approves from her phone in seconds and goes back to sleep.
The result: successful deploy, 8 hours of sleep, and zero anxiety.
Real case #2: The production bug that appeared at the gym
The situation: Diego is a freelancer in Mexico City. He’s in the middle of his weight training routine when he gets a message from his client: there’s a 500 error on the payments endpoint. Classic timing.
Before Remote Control: Diego had to rush out, get to his computer, set up the environment, find the error. Minimum 45 minutes before he could see anything.
With Remote Control: Diego already had a Claude Code session active on his Mac at home. He pulls out his phone in the locker room, opens the Claude app, connects to the existing session, and writes: “There’s a 500 error on /api/payments/checkout, check the logs from the last 2 hours and tell me what’s happening.”
Claude has full access to the filesystem, logs, and code. While Diego finishes changing, the agent has already found the problem: a nullable field that broke with the latest migration. Diego approves the fix from his phone. The client never even knew about the scare.
What you need to know before using it
Current availability: It’s in Research Preview for users on the Max plan ($100-$200/month). For users on the Pro plan ($20/month), Anthropic confirmed it will arrive “soon” but without a specific date. Team and Enterprise plans are not included in this first phase.
Real limitations (no sugar-coating):
- Your terminal has to stay open — if you close it, the session dies
- One remote session per Claude Code instance
- There’s a known bug where some users see a “Contact your administrator” error when running the command — they’re working on it
- It doesn’t work with API keys, only with authenticated Pro/Max accounts
Important security tip: The session URL generated by the system has full access to that Claude instance, including the ability to approve file changes. Treat it like a credential — don’t share it or leave it in your browser history.
How is it different from Claude Code on the web?
It’s a valid question because they might seem the same. The difference is fundamental:
| Remote Control | Claude Code Web | |
|---|---|---|
| Where does it run? | On your machine | On Anthropic’s cloud |
| Filesystem access |  Your local system |
 Virtual sandbox environment |
| MCP servers | Yours, local |
Not available |
| Use case | Continue an active session | Start from scratch anywhere |
Remote Control is for the developer who is in the middle of a complex task with their entire environment configured and wants to step away from the desk without losing anything. Claude Code Web is for starting something quick from a device where you don’t have anything installed.
The big question: Is the Max plan worth it?
If you already use Claude Code intensively and the bottleneck in your productivity is being chained to your desk, Remote Control might be the argument you were missing to justify the jump to the Max plan.
For those of us on Pro, we have to wait a bit longer. But the good news is that Anthropic confirmed it as the next step — and with the release pace they’ve been keeping over the last few months, “soon” probably won’t be that long.
Have you tried it yet? Tell us about your experience in the comments — especially if you found creative use cases we didn’t think of here. 