Claude Remote Control

Remote Control lets you start a Claude Code session in your terminal and carry on from your phone, tablet, or any browser, while execution never leaves your machine. Your filesystem, your MCP servers, and your project config stay exactly where they are.

This is an independent publication. Remote Control Handbook is written and maintained separately from Anthropic, by developers who use Claude Code daily. We check every page against the official documentation and say so when we disagree with it. Last reviewed 9 August 2026. How this site is sourced.

Just need the command? Run /rc inside a session you already have open, or claude remote-control from your shell for server mode. Both print a session URL and a QR code. Full command reference.

Where your code runs Remote Control executes on your own machine, keeping files and MCP servers local while only the conversation travels to Anthropic. Claude Code on the web executes on Anthropic infrastructure from a GitHub clone, with no access to your machine. REMOTE CONTROL CLAUDE CODE ON THE WEB Your machine Anthropic infrastructure Your filesYour MCP serversYour project configCode executes here Clone from GitHubCloud tooling onlyCloud environmentCode executes here Leaves your machine: the conversation transcript Never leaves: your code Needs from you: a pushed GitHub branch Cannot reach: anything local vs
The one difference that explains every other difference: where the code actually executes.

The problem it solves

You kick off a major refactor, Claude Code is humming along in your terminal, and then life interrupts. A meeting pulls you away. Someone needs the car. You are left choosing between abandoning the session, carrying your laptop around, or rigging up SSH tunnels and tmux just to watch progress from your phone.

The tension is real: Claude Code is powerful precisely because it runs locally, with access to your files, your MCP servers, and your tooling. That is what separates it from cloud-hosted assistants. But local has always meant tied to your desk.

Remote Control breaks that link. The session runs on your machine; the phone or browser is a window onto it.

What you get

  • Your full local environment, remotely. Filesystem, MCP servers, tools, and project configuration stay available. Typing @ autocompletes file paths from your local project.
  • Both surfaces at once. The conversation, plus subagent and workflow progress, stays in sync across every connected device. Send messages from terminal, browser, and phone interchangeably.
  • Images and files from your phone. Attachments you add in the Claude app are downloaded to your machine and passed to Claude as an @ file reference.
  • Resilience. If your laptop sleeps or the network blips, the session reconnects automatically and queued status updates are delivered once it recovers.

Four ways to start it

Where you areCommandResult
Your shellclaude remote-controlServer mode, hosting up to 32 concurrent sessions
Your shellclaude --remote-control / --rcA normal interactive session that is also remotely reachable
In a session/remote-control / /rcConverts the current session, keeping its history
VS Code/remote-control / /rcSame, with a status banner above the prompt box

Full flag list, session naming rules, and what the /rc active footer indicator means: see the command reference.

Requirements

  • Plan: available on Pro, Max, Team, and Enterprise. On Team and Enterprise it is off by default until an Owner enables the Remote Control toggle in Claude Code admin settings.
  • Login: claude.ai OAuth via /login. API keys are not supported, and neither are long-lived tokens from claude setup-token.
  • Endpoint: must talk to api.anthropic.com. Not available on Bedrock, Google Cloud Agent Platform, or Microsoft Foundry, or through a custom ANTHROPIC_BASE_URL.
  • Feature flags: DISABLE_TELEMETRY, DO_NOT_TRACK, CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC, and DISABLE_GROWTHBOOK each block it.
  • Workspace trust: run claude once in your project directory. Trust is never saved for your home directory, so start from a real project folder.

Step-by-step walkthrough on the setup page.

Remote Control is not Claude Code on the web

They sound alike and share the claude.ai/code interface, which is why they get confused. The difference is where the session runs.

Remote ControlClaude Code on the web
Runs onYour machineAnthropic cloud infrastructure
Local filesystemFull accessNo access (isolated cloud environment)
Your MCP serversAvailableCloud tooling only
Machine must stay onYesNo
Best forSteering work already underwayStarting work with no local setup

Longer comparison, including Dispatch, Channels, Slack, and scheduled tasks: Remote Control vs Claude Code on the web.

Security in one paragraph

Your local session makes outbound HTTPS requests only and never opens an inbound port. Traffic runs over TLS through the Anthropic API using short-lived, single-purpose credentials. Execution and filesystem access stay on your machine.

One correction worth making, because it is widely repeated: Remote Control is not end-to-end encrypted in the sense people usually mean. While connected, the session transcript (your messages, Claude’s responses, and tool activity) is stored on Anthropic servers. That is what keeps devices in sync and allows reconnection. Your code stays local; your conversation does not. Details on the security page.

Limitations

  • One remote session per interactive process. Use server mode to run several from one process.
  • The local process must keep running. Close the terminal or quit VS Code and the session ends. Use tmux or screen on machines you reach over SSH.
  • Extended network outage. More than roughly ten minutes offline and the session times out and exits.
  • Some commands are local-only. /plugin and /resume do not work remotely.

Practical workflows

The refactor handoff

You are deep in a refactor when the meeting reminder fires. Instead of abandoning the session, open the Claude app. During the meeting you glance at progress, approve a file change, steer Claude off an approach that will not work. Back at your desk, the refactor is three steps further along.

MCP tooling on the go

If your workflow depends on local MCP servers (a database tool, a linting server, a project-specific integration), Remote Control keeps all of it available, because execution never leaves your machine. You can trigger MCP-powered actions from your phone.

The parallel project dashboard

Run claude remote-control --spawn worktree and each on-demand session gets its own git worktree, so parallel sessions do not fight over the same files. Switch between them in the Claude app like chat threads.

Walk and code

Start a task, pick up your phone, go for a walk. Claude keeps working, and push notifications tell you when it finishes or needs a decision. Set them up on the mobile page.

Where to next

New here? Start with setup, or jump straight to the command reference if you already have a session running. This site is independent and not affiliated with Anthropic.