What Is Claude Code Remote Control and How Do You Use It?

What Is Claude Code Remote Control and How Do You Use It?

You start a long refactor at your desk, then you have to leave. The job has twenty minutes left and nobody to answer the one question it will ask.

Remote Control is the feature built for exactly that gap. It lets you pick the session up on your phone, answer the question, and let the work finish.

The part people get wrong is what it actually does. It is not Claude Code running in the cloud, and it is not a copy of your project uploaded somewhere.

It is a window into the session already running on your own machine. That distinction decides almost everything about how you set it up and what it can do.

The Short Answer

Remote Control connects claude.ai/code or the Claude mobile app to a Claude Code session running on your computer.

Anthropic's documentation is direct about where the work happens: "Claude keeps running locally the entire time, so your code execution and filesystem access stay on your machine."

You turn it on with one of three commands, get a session URL and a QR code, and open that session from any browser or from the Claude app on your phone.

Your machine has to stay awake and the claude process has to keep running. Close the laptop lid and the window goes dark.

Getting found in AI search is its own problem. See how your site performs when people ask an assistant instead of a search engine. Get a free SEO audit.

What Remote Control Actually Gives You

Four things, and all four follow from the session being local rather than hosted.

Your real environment, remotely. Your filesystem, your MCP servers, your tools and your project configuration all stay available, and typing @ still autocompletes paths from your local project.

Both surfaces at once. The conversation and the progress of subagents and workflows stay in sync across every connected device, so you can type from the terminal, the browser or the phone interchangeably.

Files and photos from your phone. Attach a photo and Claude sees the image directly as part of your message. Other files get downloaded to your machine and passed in as @ references.

Recovery from a dropped connection. If your laptop sleeps or the network drops, Claude Code reconnects automatically once the machine is back online.

How It Differs From Cloud Sessions

This is the comparison worth being clear about, because the two features look similar in the session list.

Cloud sessions run on Anthropic's infrastructure. They do not need your machine and they do not touch your local files.

Remote Control sessions run on your machine and interact with your local filesystem. The web and mobile interfaces are, in the documentation's own words, "a window into that local session."

So cloud sessions are the right answer when your computer is off, and Remote Control is the right answer when the work needs your actual repository, credentials and tooling.

What Do You Need Before It Will Work?

More than you might expect, and most setup failures trace back to this list rather than to the commands.

A qualifying plan. Remote Control is available on Pro, Max, Team and Enterprise. API keys are not supported at all.

An admin toggle, on Team and Enterprise. An Owner has to enable the Remote Control toggle in Claude Code admin settings before anyone in the organisation can use it.

A claude.ai login. Run claude and sign in with /login. Without an eligible login the server command exits with an error.

The standard API endpoint. Remote Control is unavailable if you route through Amazon Bedrock, Google Cloud's Agent Platform or Microsoft Foundry.

It is also unavailable if ANTHROPIC_BASE_URL points at anything other than Anthropic's own API, which includes most LLM gateways and proxies. Unset the variable to use the feature.

Feature-flag traffic left on. Setting CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC or DISABLE_GROWTHBOOK makes Remote Control unavailable entirely.

Setting only DISABLE_TELEMETRY or DO_NOT_TRACK is less severe: the feature stays available unless your organisation requires Trusted Devices, and it needs a recent Claude Code build.

A trusted workspace. In a directory you have not trusted, the command explains what trusting it enables and asks before starting.

How Do You Start a Remote Control Session?

There are three entry points, and picking the right one is most of the decision.

1. Server mode, for one machine serving many sessions. Run claude remote-control in your project directory. The process stays in the foreground waiting for connections.
2. An interactive session with it enabled. Run claude --remote-control, or --rc for short. You get a normal terminal session that is also reachable remotely, so you can still type locally.
3. From a session already in progress. Type /remote-control, or /rc. This carries your existing conversation history across, which is what you want when you are leaving mid-task.

The first time you use any of them, Claude Code asks for a one-time confirmation before it starts. Decline and it simply asks again next time.

Server mode is the one with the most knobs. --name "My Project" sets a session title you will recognise in the list, and --capacity caps concurrent sessions at something below the default of 32.

--spawn worktree gives each on-demand session its own git worktree, which is how you avoid two remote sessions editing the same files. --spawn session serves exactly one and rejects everything else.

--permission-mode sets the starting permission mode, and --chrome or --no-chrome decides whether those sessions can drive Chrome on your machine while you work from elsewhere.

Your docs rank. Does your product page? A free audit shows which of your pages search engines and AI assistants actually surface. Request yours.

Connecting From the Other Device

Once a session is live there are three ways in, and they all land in the same place.

Open the session URL in any browser to go straight to it on claude.ai/code. In server mode the URL is printed in your terminal.

Scan the QR code to open it in the Claude app instead. In server mode, press the spacebar to toggle the QR display on and off.

Or open the Claude app, tap Code in the navigation, and find the session by name. Remote Control sessions are marked with a computer icon and a green dot when they are online.

If you do not have the app yet, running /mobile inside Claude Code shows a QR code that opens the right app store for your phone.

What the Connected Device Can Do

More than read along. The device shows any subagents and workflows already running, and stopping one there stops that task on your machine.

You can pick the model from the device and the session on your machine switches to it. The same applies to the effort level, unless you have pinned one with an environment variable.

On a git repository, the device's diff pane shows your changes, including uncommitted edits, measured against the repository's default branch.

Compaction and /clear are reflected too: connected devices show compaction progress and reset alongside your terminal.

Turning It On For Every Session

Remote Control only activates when you ask for it, unless you enable auto-connect.

Run /config inside Claude Code and set Enable Remote Control for all sessions to true, and every interactive session will connect on startup.

The Desktop app has the same control under Settings then Claude Code, and the VS Code extension has it in the command menu's settings section.

From a settings file, remoteControlAtStartup does the same job in your user settings or in managed settings.

There is a deliberate asymmetry worth knowing: project and local settings files can turn auto-connect off, but a true in a checked-in project file is ignored.

That means nobody can commit a file that silently enables Remote Control for everyone who clones the repository. Auto-connect also signs in as you, so a session it starts appears only in your own account.

Shipping faster than your site ranks? A free audit shows the gap between what you build and what search surfaces. Get a free audit.

Limitations and Risks Worth Knowing

The design is sound, and the sharp edges are all consequences of it rather than bugs.

Your machine is the dependency. It must stay on, awake and running the process. Remote Control is not a way to work while your laptop is shut.

Stopping the server has a four-hour grace period. After Ctrl+C you can bring sessions back with claude remote-control, --continue or --session-id, but only for about four hours.

/resume and /teleport do not carry history to the device. Switching conversations locally means the connected device does not receive the earlier history, though new messages flow both ways.

Running two terminals needs care. Resume a conversation in a second terminal while the first still has Remote Control on and the second prints a notice and stays off, rather than stealing the session.

It is a real remote control. Anyone with access to your signed-in Claude app or browser session can drive a process with your filesystem and your credentials. Treat phone access as you would an unlocked terminal.

--sandbox was removed. Passing --sandbox or --no-sandbox to the server now makes it exit with an error. Sandboxing is configured in a settings file instead.

On the security side, the architecture is reassuring: your local session makes outbound HTTPS requests only and never opens inbound ports on your machine.

It registers with the Anthropic API and polls for work, so there is no listening socket on your laptop for anyone on the same network to find.

Before you ship, check what search sees. One free audit, one prioritised list of fixes for your site. Start the audit.

Remote Control vs the Alternatives

Three options overlap here, and they solve genuinely different problems.

Remote Control is for continuing work that needs your machine: your repository state, your local services, your MCP servers, your credentials.

Cloud sessions are for starting work when your machine is unavailable. They run on Anthropic's infrastructure and have no access to your local files.

SSH into your own box is the old answer and still works, but you lose the synced conversation, the diff pane, the mobile interface and the automatic reconnection.

The honest test is whether the task needs your local environment. If it does, Remote Control. If it does not, a cloud session is simpler and does not care whether your laptop is open.

Final Thoughts

Remote Control is best understood as a mobile interface to a process you already trust, not as a new place for that process to live.

Set up the one-time pieces first, since that is where the time goes: a qualifying plan, the admin toggle if you are on a team, a claude.ai login, and no gateway in the way.

Then pick the entry point by situation. /remote-control when you are leaving mid-task, claude --remote-control when you want to work from both ends, and server mode when one machine should serve several sessions.

If you are newer to the terminal side of all this, our guide to the Claude Code CLI covers installation, updates and the flags worth knowing before you start handing sessions to your phone.

The feature earns its place the first time a build finishes correctly because you answered one question from a train platform.