AI Catchup

Claude Code Remote Control Can Start Sessions From Your Phone

By 5 min read

Claude Code's Remote Control update lets you start a session from the Claude mobile app, choose a directory on a connected machine, and continue working from a phone, tablet, or browser. The code and tools still run on the local machine, while the phone acts as a remote control surface.

Claude Code's Remote Control can now start a session from your phone. Anthropic says a machine running claude remote-control appears as a device card in the Claude app's Code tab. Tap the machine, choose a directory, and the session starts on that computer. (ClaudeDevs on X)

The phone is a control surface, not the execution environment. Claude Code keeps running on the local machine, with local filesystem access, tools, MCP servers, and project configuration staying there while the conversation and progress are synchronized to the phone or browser. (Claude Code Remote Control documentation)

Start from the Claude app

The documented flow is:

  1. In a project directory on the machine, run claude remote-control.
  2. Open the Claude app and tap Code.
  3. Find the machine's device card at the top of the session list.
  4. Tap the machine and choose a directory.
  5. Start the Claude Code session on that machine, then continue from the phone.

The X announcement describes the phone-start flow, while the documentation lists claude remote-control as server mode and explains that the session URL or QR code can open the session in claude.ai/code or the Claude app. (ClaudeDevs on X; Claude Code Remote Control documentation)

You can also expose an existing session instead of starting a new one. The documentation lists --remote-control and --rc for an interactive session, plus /remote-control or /rc inside an existing session. An existing session carries over its current conversation history when Remote Control is enabled. (Claude Code Remote Control documentation)

What stays on the computer

Remote Control does not move the project to the phone. Claude Code keeps code execution and filesystem access on the local machine, and the documentation says the local filesystem, MCP servers, tools, and project configuration are made available remotely through the connected session. (Claude Code Remote Control documentation)

That boundary makes the feature useful for long-running work without turning a phone into a development workstation. A session can keep running in a terminal or server process while you inspect progress, answer a question, send the next prompt, or attach a photo from the Claude app. Files attached from the app are downloaded to the machine and passed to Claude as file references. (Claude Code Remote Control documentation)

The phone can also show subagents and workflows already running in the background. Stopping one from the phone stops that task on the machine, so treat the connected device as an active control surface rather than a read-only dashboard. (Claude Code Remote Control documentation)

Sync and reconnection improvements

The approved ClaudeDevs thread pairs the phone-start workflow with synchronization and reconnect changes. Anthropic says the phone stays in sync with what is running on the machine, resuming on a laptop keeps the phone on the live session instead of archiving it, and the phone shows the session offline shortly after Claude Code exits. (ClaudeDevs on X)

The durable documentation describes the separate reliability behavior in more detail. If the laptop sleeps or the network drops, Claude Code reconnects automatically when the machine returns online, and queued status updates from subagents and workflows are delivered after recovery. The local process still has to remain alive for the session to stay available. (Claude Code Remote Control documentation)

This distinction matters: reconnect behavior helps an active session survive an interruption, while the phone-start feature is what lets you begin a new session on a connected machine without first sitting at that machine. The two capabilities work together, but they solve different problems.

Plans and setup boundaries

Anthropic's documentation lists Remote Control for Pro, Max, Team, and Enterprise plans. Team and Enterprise workspaces have an extra administrative gate: an Owner must enable the Remote Control toggle in Claude Code admin settings. (Claude Code Remote Control documentation)

Remote Control does not use an Anthropic API key. The documentation says API-key authentication is unsupported and that the feature requires a claude.ai login. It is also unavailable when using Amazon Bedrock, Google Cloud's Agent Platform, or Microsoft Foundry, and unavailable when ANTHROPIC_BASE_URL points to a host other than api.anthropic.com. (Claude Code Remote Control documentation)

Before trying it, run Claude Code from the project directory at least once to accept the workspace trust dialog. Keep the machine awake and connected for the workflow you want to control remotely. If the process exits, the connected session goes offline rather than continuing to execute work on the phone. (Claude Code Remote Control documentation)

A useful first workflow

Start with a bounded task on a machine you control: run claude remote-control in a repository, pick the project directory from the phone, and ask Claude Code to inspect or test a small change. Keep the first session visible on the machine so you can confirm permissions, MCP connections, and the project context before relying on phone-only control.

For longer work, use the phone to check progress and answer prompts, but keep the local execution boundary in mind. The session remains tied to the machine's filesystem and credentials, so anyone who can control the connected session should be treated as having meaningful access to that environment. The official documentation confirms the connection model and prerequisites; it does not make a general security guarantee for every project setup. (Claude Code Remote Control documentation)

Sources

Keep building the workspace playbook

Frequently Asked Questions

Can Claude Code run directly on a phone?

No. Remote Control connects a Claude Code session running on a local machine to the Claude app or claude.ai/code. The machine keeps running the code, tools, and filesystem work.

How do you start a Claude Code session from a phone?

Run claude remote-control on a machine, then open the Code tab in the Claude app. The connected machine appears as a device card; tap it, choose a directory, and start the session on that machine.

Which plans support Claude Code Remote Control?

Anthropic's documentation lists Pro, Max, Team, and Enterprise plans. On Team and Enterprise, an Owner must enable the Remote Control toggle in Claude Code admin settings.

Does Remote Control work with an Anthropic API key?

No. Anthropic's documentation says API-key authentication is not supported for Remote Control. The feature requires signing in through claude.ai, and it is unavailable with the listed cloud-provider credentials and custom API base URLs.

Get the weekly AI Catchup

Tools, practices, and what matters, in your inbox every week.