Claude Code Remote Control: Drive Your Local Dev Session From Any Device

What Remote Control actually is

Claude Code Remote Control is a feature from Anthropic that connects `claude.ai/code` or the Claude mobile app to a Claude Code session running on your own machine. The idea is straightforward: start a coding task at your desk, then keep working on it from your phone on the couch, or from a browser on a different computer.

The important detail is where the work happens. Unlike Claude Code on the web, which runs on cloud infrastructure, a Remote Control session runs directly on your machine and touches your local filesystem. The phone and browser interfaces are just windows into that local session. Nothing moves to the cloud — Claude keeps executing locally the whole time.

That means your full local environment stays available remotely: your filesystem, your MCP servers, your project configuration, and your tools. Typing `@` autocompletes file paths from your local project even when you're issuing commands from a browser. The conversation stays in sync across every connected device, so you can send messages from your terminal, browser, and phone interchangeably. If your laptop sleeps or the network drops, the session reconnects when the machine comes back online.

Remote Control is in research preview and requires Claude Code v2.1.51 or later. Check with `claude --version`.

How you start and connect a session

You can launch a session from the CLI or the VS Code extension. From the terminal, `claude remote-control` starts the process in server mode: it stays running, waits for remote connections, and prints a session URL. Press spacebar to show a QR code for quick access from your phone. While a remote session is active, the terminal shows connection status and tool activity.

There are a few useful flags. `--name "My Project"` sets a custom session title visible in the session list at `claude.ai/code`. `--spawn worktree` gives each on-demand session its own git worktree so sessions don't clobber each other by editing the same files — the default `same-dir` shares the working directory and can conflict.

The bookmarked context also mentions `/teleport` and `claude --teleport`, which move a cloud session down to your machine to continue it in the terminal — the reverse direction. And you can flip on "Enable Remote Control for all sessions" in `/config` if you'd rather not opt in each time.

A few requirements are worth flagging before you get excited:

• It works on Pro, Max, Team, and Enterprise plans. API keys are not supported.
• On Team and Enterprise, an Owner must enable the Remote Control toggle in Claude Code admin settings first — it's off by default.
• It is not available on Amazon Bedrock, Google Vertex AI, or Microsoft Foundry. As of v2.1.196 it's also disabled when `ANTHROPIC_BASE_URL` points at anything other than `api.anthropic.com`, such as an LLM gateway or proxy.
• You need to have run `claude` in the project directory at least once to accept the workspace trust dialog.

That last set of constraints matters for teams. If you route Claude through a proxy or gateway — a common setup for cost tracking and governance — Remote Control simply won't work until you unset that variable. This is a consumer-and-subscription feature, not something you bolt onto an API-driven pipeline.

Where it fits for automation-minded builders

Be clear-eyed about what this is. Remote Control is not a headless agent orchestration layer, and it's not a replacement for scheduled or event-driven automation. It's a way to *stay in the loop* on a session that's already running on hardware you control. If your automation instinct is to fire off agents unattended, this isn't that tool. For that you'd reach for the Claude Code Agent SDK, or wire model calls into an orchestrator like n8n, Make, or Zapier.

Where Remote Control earns its keep is the messy middle of real development work: long-running tasks that need occasional human judgment. You kick off a refactor or a build at your desk, walk away, and approve or redirect from your phone when Claude hits a decision point. The session survives sleep and network drops, so you're not babysitting a terminal. For solo developers and small teams who live in Claude Code, that's a genuinely practical convenience — you get cloud-like reach without giving up your local filesystem, MCP servers, and secrets.

The honest comparison is with Claude Code on the web. The web version runs in the cloud and is the right choice when you don't need local files or tools. Remote Control is the right choice when the work has to touch your machine — your repos, your local databases, your configured MCP integrations. Same interfaces, very different execution model. On Team and Enterprise, the Trusted Devices option adds an enrollment layer if your organization wants to control which devices can connect, which is a reasonable guardrail given that a remote window is effectively driving a session with access to your local environment.

If you want help wiring Claude Code into your development and automation workflows, browse the provider directory to find people who can put it to work.