
Claude Code remote control is a feature that enables developers to keep a local Claude Code session running on their machine and pick it up seamlessly from a phone, tablet, or any browser. Designed for anyone who codes with Claude on the command line or in VS Code, it bridges the gap between desk-based work and on-the-go access. The core value is uninterrupted development flow: you start a task at your terminal, then travel or step away, and continue exactly where you left off using the Claude mobile app or the web interface at claude.ai/code. The session stays alive on your machine, so all your local files, tools, and server connections remain available as if you never left.
A major pain point for developers is being chained to a single workstation. Before remote control, stepping away from the terminal meant abandoning a complex debugging session or waiting until you returned to approve a suggestion. Now, you can send messages from your phone while lounging on the couch or monitor progress from a laptop in another room. This matters because coding often involves long-running tasks—tests, builds, code generation—that shouldn’t require constant presence. Claude Code remote control decouples you from the local keyboard, making coding truly portable and responsive to your life.
The first standout feature is multi-surface connectivity. Using the `claude remote-control` command, you start a server process that registers your session with Anthropic’s API and then provides a session URL and a QR code. You can open that URL in any browser or scan the QR with the Claude mobile app to immediately join the session. Once connected, you type commands from the browser, and they execute instantly on your local machine, with file paths auto-completed from your project. The session appears in a list at claude.ai/code, marked by a computer icon and a green dot, so you can find it easily even if you forget the URL.
Remote control preserves your entire local development environment. Your filesystem, MCP servers, installed tools, and project configuration stay active because Claude continues to run locally. When you type `@` in the web or mobile interface, it autocompletes file paths from your project directory on your machine, giving you the same rich context you would have at the terminal. This is fundamentally different from web-only versions of Claude Code that run in the cloud; here, your MCP server plugins and custom tool integrations work unchanged, and you never sacrifice access to proprietary code or local databases.
admin
Another crucial capability is interruption survival and session management. If your laptop goes to sleep or your network drops, the session automatically reconnects when your machine wakes up or comes back online, without losing conversation history. For more advanced setups, the `--spawn` flag enables server mode where you can run multiple concurrent sessions, each either sharing the same directory or getting its own git worktree for isolation. The `--capacity` option lets you set a maximum number of sessions, up to 32 by default. You can even set custom session names with `--name` to keep multiple projects organized.
Remote control works through a secure, outbound-only architecture. Your local Claude Code process initiates an HTTPS connection to the Anthropic API and polls for incoming messages; no inbound ports are ever opened on your machine. When you connect from another device, the API routes messages between the web client and your local session over a streaming connection encrypted via TLS. Short-lived, single-purpose credentials secure each transaction, expiring independently. The system also supports managed settings: on Team and Enterprise plans, admins can toggle the feature on or off, and IT can disable it on specific devices via configuration files, ensuring compliance with organizational policies.
Real-world use cases are diverse. A developer starts a large refactoring in a terminal before dinner, then joins the session from an iPad while relaxing, approving or rejecting Claude’s changes one by one. A data engineer kicks off a batch processing script, walks to a coffee shop, and receives a mobile push notification when the job completes, with a prompt to review the output. In a team environment, a lead might run a server-mode instance on a shared build machine, allowing team members to connect and query logs or trigger deployments without SSH. Another user uses VS Code’s `/remote-control` command and sends a message from a phone while the editor is closed, then reopens it later to find all changes applied.
Target users span individual developers on Pro and Max plans to organizations on Team and Enterprise. It supports command-line workflows (any terminal running Claude Code v2.1.51+), the VS Code extension (v2.1.79+), and the Claude mobile apps for iOS and Android. Administrators can selectively enable remote control for their teams via the admin settings panel. Mobile push notifications and the `CLAUDE_CLIENT_PRESENCE_FILE` option further tailor the experience for asynchronous collaboration. Ultimately, Claude Code remote control makes your coding assistant portable, secure, and deeply integrated with your local toolchain—ensuring you never have to stop coding just because you’re away from your desk.
Professional developers and teams who use Claude Code for local coding workflows. Ideal for individuals on Pro or Max plans who need mobile flexibility, and for Team and Enterprise organizations that want to enable remote access while maintaining control via admin policies. Users must be comfortable with CLI tools and optionally use the VS Code extension. The solution is suited for those who rely on MCP servers, custom scripts, and local environment configurations that cannot be moved to the cloud. It is not for casual web‑only users but rather for engineers who want to extend their existing terminal‑based Claude Code sessions to mobile devices and alternate workstations without sacrificing the power of their local toolchain.
Updated 2026-02-28