Anthropic just made it possible to control your AI coding agent from your phone. On February 25, 2026, the company launched Claude Code Remote Control — a new feature that lets you start a coding session on your desktop, then pick it up from any browser or the Claude mobile app on iOS and Android. Your code stays local. Your tools stay local. But now you can monitor, prompt, and steer Claude Code from anywhere. Here's everything you need to know.
What Is Claude Code Remote Control?
Claude Code Remote Control is a new feature that connects claude.ai/code or the Claude mobile app (iOS/Android) to a Claude Code session running on your local machine. Think of it as a remote window into your terminal — except it's a full conversational interface with all of Claude's capabilities.
Here's the key concept: nothing moves to the cloud. Your project files, MCP servers, tools, and configuration all stay on your local machine. The web and mobile interfaces are just a window into that local session. Claude keeps running on your hardware, accessing your filesystem, your environment, your setup. You're simply viewing and controlling it from somewhere else.
This solves a real problem. Before Remote Control, if you started a long Claude Code task — say, refactoring a module or writing tests across a codebase — you were tethered to your terminal. Step away from your desk? You'd have to check back later. Now you can kick off a task, grab your phone, walk to a coffee shop, and keep directing Claude from there.
The conversation stays in sync across all connected devices. You can type a prompt in the terminal, see the response on your phone, then follow up from your browser. Terminal, browser, phone — they're all interchangeable windows into the same session.
Remote Control also auto-reconnects if your laptop sleeps or your network drops briefly. If you close your laptop lid and open it ten minutes later, the session picks back up. This is a Research Preview that launched on February 25, 2026, available now for Max plan users and coming soon to Pro users.
How to Set Up Claude Code Remote Control
Getting started takes about two minutes. Here's the step-by-step process.
Step 1: Make Sure You're Signed In
Remote Control requires a Pro ($20/mo) or Max (from $100/mo) subscription. It does not work with API keys. If you haven't already, run Claude Code and sign in:
claude
/loginThis opens a browser window to authenticate via claude.ai. Complete the sign-in flow.
Step 2: Trust Your Workspace
Before remote control will work, you need to accept the workspace trust dialog. Navigate to your project directory and run Claude Code once:
cd /path/to/your/project
claudeAccept the trust dialog when prompted. This is a one-time step per project directory.
Step 3: Start Remote Control
You have two options:
Option A: Start a new session with remote control enabled:
claude remote-controlOption B: Enable remote control from an existing session:
/remote-controlOr the shorthand:
/rcBoth approaches display a session URL and a QR code (press spacebar to show the QR code in your terminal).
Step 4: Optional Flags
The claude remote-control command supports a few useful flags:
--verbose— Show detailed connection logging--sandbox/--no-sandbox— Control sandbox mode for the session
Step 5: Enable for All Sessions (Optional)
If you want every Claude Code session to automatically be available for remote control, run:
/configThen select "Enable Remote Control for all sessions". This saves you from typing /rc every time.
Connecting From Another Device
Once remote control is active, you have three ways to connect from another device:
Method 1: Session URL
Copy the session URL displayed in your terminal and open it in any browser on any device. It routes to claude.ai/code and connects to your local session automatically.
Method 2: QR Code
Press spacebar in the terminal to display a QR code. Scan it with your phone's camera — it opens directly in the Claude mobile app (iOS or Android). This is the fastest way to get connected on mobile.
Method 3: Claude App / claude.ai
Open the Claude app on your phone or navigate to claude.ai/code in a browser. Your active remote session appears in the session list with a computer icon and a green dot indicating it's online. Tap it to connect.
All three methods give you the same full interface. You can switch between them freely — the conversation stays in sync regardless of which device you're using at any given moment.
Security Model: How It Works Under the Hood
Security is the first question most developers ask when they hear "control your local machine from your phone." Here's how Anthropic designed the security model for Remote Control.
Outbound HTTPS only. Remote Control never opens inbound ports on your machine. Your computer doesn't become a server. Instead, it registers with the Anthropic API and polls for work — the same way your browser polls for new emails. All communication flows outward from your machine over TLS through Anthropic's API.
Multiple short-lived credentials. Rather than using a single long-lived token, the system uses multiple short-lived credentials, each scoped to a single purpose. These credentials expire independently, which limits the blast radius if any single credential were compromised.
No inbound attack surface. Because your machine only makes outbound HTTPS connections, there's no port to scan, no service to exploit, no firewall rule to configure. This is a fundamentally different security posture than running a web server or SSH tunnel.
It's worth noting that Remote Control does not appear to support the --dangerously-skip-permissions flag, which aligns with the cautious security approach. When you're controlling a coding agent remotely, enforcing the permission model is especially important since you may not be watching every action in real time.
Remote Control vs. Claude Code on the Web
Anthropic now offers two ways to use Claude Code outside your terminal: Remote Control and Claude Code on the Web. They serve different purposes.
| Feature | Remote Control | Claude Code on the Web |
|---|---|---|
| Where code runs | Your local machine | Anthropic's cloud infrastructure |
| Local files accessible | Yes — full filesystem | No — cloud sandbox only |
| MCP servers | Your local MCP servers available | Not available |
| Project config | Uses your local .claude config | Separate cloud config |
| Setup required | Local Claude Code installation + login | None — runs in browser |
| Best for | Continuing local work from another device | Tasks without local setup, repos not cloned, parallel work |
| Parallel sessions | One session at a time | Multiple sessions possible |
| Requires terminal open | Yes | No |
The simple rule: Use Remote Control when you want to keep working with your local environment from another device. Use Claude Code on the Web when you want a clean cloud session without any local dependencies.
Limitations and Gotchas
Remote Control is a Research Preview, and it comes with constraints you should know about before relying on it:
- One session at a time. You can only have one remote control session per Claude Code instance. You can't remote-control multiple terminal sessions simultaneously.
- Terminal must stay open. Claude Code is still a local process. If you close the terminal window or kill the process, the remote session dies. Your laptop needs to be on and the terminal running.
- Network timeout (~10 minutes). If your machine loses internet connectivity for an extended period — roughly 10 minutes — the session times out. You'll need to restart remote control when connectivity returns.
- No
--dangerously-skip-permissionsflag. Remote sessions enforce the standard permission model. You'll need to approve sensitive actions. - Not available for Team/Enterprise. Currently limited to individual Pro ($20/mo) and Max (from $100/mo) plans. Team and Enterprise plans are not supported yet.
- No API key support. You must authenticate via claude.ai subscription. API keys won't work.
Also Announced: Cowork Scheduled Tasks
On the same day as Remote Control, Anthropic also launched scheduled recurring tasks in Cowork — Claude Code's general-purpose agent sibling that runs in Claude Desktop.
Cowork scheduled tasks let you set up recurring work that Claude handles automatically. Think daily code reviews, weekly dependency checks, or regular test suite runs. It's the automation layer that complements Remote Control's on-demand mobile access.
There's one important caveat: scheduled tasks only run while your computer is awake and the Claude Desktop app is open. If your machine is asleep when a task is scheduled, it skips that run and executes when the computer wakes up. This means it's not a replacement for CI/CD or cron jobs on a server — it's more like a smart assistant that works while you work.
Together, Remote Control and Cowork scheduled tasks represent Anthropic's push to make Claude Code useful beyond the terminal. One gives you mobile access on demand; the other gives you background automation.
What This Means for Developers
Claude Code Remote Control isn't just a convenience feature — it signals a shift in how we interact with AI coding agents.
The terminal is no longer the bottleneck. Until now, AI coding agents were desktop-bound. You sat at your machine, typed prompts, watched output, approved actions. Remote Control breaks that pattern. You can kick off a refactoring task before lunch, check progress from your phone while eating, and approve the final changes from your tablet on the couch.
Mobile-first AI coding becomes real. The Claude iOS and Android apps aren't just chat interfaces anymore — they're legitimate development tools when connected to a Remote Control session. You have full access to your local filesystem, your MCP servers, your project configuration. The phone becomes a real development interface, not a toy.
The always-on coding agent is getting closer. Remote Control plus Cowork scheduled tasks paint a picture: an AI agent that works in the background, checks in with you on mobile, and handles routine tasks on a schedule. We're moving from "AI assistant you talk to at your desk" to "AI teammate that works alongside you everywhere."
For developers already using Claude Code, Remote Control is an immediate productivity unlock. For the broader industry, it's a signal that AI coding tools are evolving beyond the IDE plugin model into something more ambient and always-available.
How Serenities AI Helps You Build Faster
If you're excited about Claude Code Remote Control, you're likely already building with AI tools daily. Serenities AI is an all-in-one platform that brings together AI coding, content, and automation in one workspace — so you don't need to juggle a dozen different tools.
Serenities AI integrates with Claude and other AI models, giving you a unified interface for development workflows. Whether you're using Claude Code for coding, Claude for content, or other AI tools for automation, Serenities AI acts as your command center.
Plans start at Free with paid tiers at $24/mo, $49/mo, $99/mo, and $249/mo — so you can start without commitment and scale as your needs grow. Check out Serenities AI to see how it fits into your AI-powered development workflow.
FAQ
Does Claude Code Remote Control require a paid plan?
Yes. Remote Control requires a Pro ($20/mo) or Max (from $100/mo) Anthropic subscription. It does not work with API keys. Currently, it's available as a Research Preview for Max users and is coming soon to Pro users. Team and Enterprise plans are not supported yet.
Can I use Remote Control from my phone?
Absolutely. You can connect via the Claude app on iOS or Android by scanning the QR code displayed in your terminal, or by opening the app and selecting the active session (look for the computer icon with a green dot). You can also use any mobile browser by navigating to the session URL.
Does my computer need to stay on?
Yes. Claude Code runs as a local process in your terminal. The terminal must stay open and your machine must remain powered on. If your laptop goes to sleep briefly, the session auto-reconnects when it wakes. However, an extended network outage of roughly 10 minutes will cause the session to time out, and you'll need to restart remote control.
Is Remote Control the same as Claude Code on the Web?
No. Remote Control connects to a Claude Code session running on your local machine — giving you access to your filesystem, MCP servers, and project configuration. Claude Code on the Web runs on Anthropic's cloud infrastructure with no access to your local environment. Use Remote Control to continue local work from another device; use the Web version for standalone cloud sessions.
Bottom Line
Claude Code Remote Control turns your phone into a legitimate development interface. For the first time, you can control a local AI coding agent from anywhere — your couch, a coffee shop, or your commute. It's a Research Preview with some limitations, but the foundation is solid: outbound-only security, seamless sync across devices, and full access to your local environment.
If you're building with AI tools daily, try Serenities AI free to bring all your AI workflows together in one platform. And if you're a Claude Max subscriber, go run claude remote-control right now — your phone is waiting.