Claude Remote Control: How It Works, Turning It Off, and Auto Mode

Once you start using Claude Code Remote Control, the same handful of questions come up: what is actually happening between my laptop and my phone, how do I make it stop, and why can't I pick auto mode from the app? This is the practical Q&A follow-up to our Claude Code Remote Control guide, which covers the basics and the overall workflow. Everything below is checked against Anthropic's current Claude Code documentation.

What is Remote Control in Claude?

In one sentence: Remote Control connects claude.ai/code or the Claude app on iOS and Android to a Claude Code session that is running on your own machine. The phone or browser is a window into that local session. Claude keeps executing locally the whole time, so your filesystem, MCP servers, tools, and project configuration all stay available — and your computer has to stay on, with the claude process still running.

That makes it different from cloud sessions, which run on cloud infrastructure. Remote Control is for steering work you already started locally from another device.

How does Claude Remote Control work?

The security model is the part most people ask about, so here it is precisely:

• Outbound only. Your local session makes outbound HTTPS requests and never opens inbound ports on your machine.
• Register, then poll. When you start Remote Control, Claude Code registers with the Anthropic API and polls for work. When you connect from another device, the server routes messages between that client and your local session over a streaming connection.
• TLS and short-lived credentials. All traffic goes through the Anthropic API over TLS, using several short-lived credentials that are each scoped to a single purpose.
• The transcript is stored; execution is not. While Remote Control is connected, the session transcript (your messages, Claude's responses, and tool activity) is stored on Anthropic's servers so devices stay in sync and can reconnect. Code execution and file access stay on your machine.

There are three ways to start it. claude remote-control runs a server mode that waits for remote connections and shows a session URL (press spacebar for a QR code). claude --remote-control (or --rc) starts a normal interactive session that is also reachable remotely. And /remote-control (or /rc) inside an existing session connects it, carrying the conversation history over. The Desktop app and the VS Code extension use the /remote-control command.

It needs a Pro, Max, Team, or Enterprise plan and a claude.ai login; API keys aren't supported. On Team and Enterprise, an Owner has to switch it on in the Claude Code admin settings first. It also won't run on Amazon Bedrock, Google Cloud, or Microsoft Foundry, or when ANTHROPIC_BASE_URL points somewhere other than api.anthropic.com.

How do I turn off Remote Control in Claude Code?

"Off" means different things depending on what you want. Pick the level that matches:

1. Disconnect one session but keep working locally

In an interactive session, run /remote-control again. That opens the status panel (session URL and QR code), and the panel lets you disconnect Remote Control while your local session keeps running. In the Desktop app and VS Code, running /remote-control a second time disconnects too.

2. Stop a server-mode session

If you started claude remote-control, press Ctrl+C. The sessions it was serving stop responding on your phone and in the browser. They aren't archived, though: for about four hours, running claude remote-control in the same directory brings them all back, and claude remote-control --continue brings back just the one it started with.

3. Stop it connecting automatically

If Remote Control turns on by itself every time you start Claude, auto-connect is enabled. Run /config and set Enable Remote Control for all sessions to false. The same toggle lives in the Desktop app under Settings > Claude Code > Connect new sessions to Remote Control, and in VS Code's command menu. In a settings file, the key is remoteControlAtStartup. A false in a project's .claude/settings.json turns auto-connect off for that repository, while a true there is ignored — so a checked-in file can never switch it on for everyone.

4. Disable it entirely

To turn Remote Control off everywhere it can start, set disableRemoteControl in a settings file:

{
  "disableRemoteControl": true
}

IT teams can push the same key through managed settings, and on Team and Enterprise plans the organization-level Remote Control toggle is off unless an Owner enables it.

Can I use Claude Remote Control with auto mode?

Yes, with one catch that confuses a lot of people. From the app or claude.ai/code, a Remote Control session you started yourself offers Manual, Accept edits, and Plan in the mode dropdown. You can't select Auto (or Bypass permissions) from the app.

But the dropdown reflects whatever mode the local session is already in, including one you set from the terminal. So the move is to choose auto mode on the machine, then connect:

• On Claude Code v2.1.283 or later, auto mode is already the built-in starting mode for interactive terminal sessions, unless your settings choose something else.
• Otherwise, press Shift+Tab until the status bar shows ⏵⏵ auto mode on, or start with claude --permission-mode auto.
• Then run /remote-control. Your phone now steers a session where a classifier reviews actions instead of asking you about each one.

For server mode, claude remote-control --permission-mode acceptEdits sets the starting mode for the sessions the server creates. If you're new to the classifier itself, our Claude Code auto mode post explains what it allows and blocks.

How do permission prompts show up on my phone?

Permission prompts appear in claude.ai and the Claude app for you to approve, and they stay open until you answer — Claude Code doesn't time them out. The same goes for questions Claude asks you directly. Other dialogs forwarded to the remote side, such as the model-choice prompt after a safety refusal, close after five minutes by default and take their no-action default; the dialogExpiry setting changes that.

Auto mode doesn't make prompts vanish completely either. Explicit ask rules still force a prompt, and if the classifier blocks actions three times in a row or twenty times in a session, auto mode pauses and Claude Code goes back to prompting until you approve one. On Remote Control, those prompts land on your phone — which is exactly when you want to know about them quickly.

Will I be told when Claude needs me?

Remote Control can send mobile push notifications. Run /config and enable Push when Claude decides (typically when a long task finishes), Push when actions required (permission prompts and questions), or both. They only work while Remote Control is active, and Claude Code skips them while you're focused on the connected terminal.

That covers the away-from-desk case. At the desk — with several sessions open, Remote Control off, or Claude working in a terminal tab you've buried — you still need something that taps you on the shoulder. We compare the options in our Claude Code notifications guide.

Quick reference

• Start: claude remote-control, claude --remote-control, or /remote-control
• Disconnect, keep working: run /remote-control again
• Stop server mode: Ctrl+C
• No auto-connect: /config → Enable Remote Control for all sessions → false
• Off everywhere: "disableRemoteControl": true
• Auto mode: set it locally, then connect — the app can't select it