Claude Code from your phone with Remote Control

The map: where each one runs

As of October 2026, from Anthropic’s docs:

Cloud sessions: nothing of yours to keep running

If your project lives on GitHub, this is the least setup: start a task from the Code tab, or with claude --cloud from a terminal, and it runs on an Anthropic machine against a clone of your repository. It “keeps running after you close your laptop,” and there’s no separate compute charge: cloud sessions share your plan’s limits. When you want the work back in a real terminal, claude --teleport pulls it down. The handoff is one-way from the CLI (only the desktop app can send a local session to the cloud), and after a teleport the terminal has its own copy, so to keep steering it from your phone, start /remote-control in it. A cloud session that sits idle stops, and its machine is reclaimed.

Remote Control: your computer, steered from the phone

This is the one most people picturing “Claude Code from my phone” want: the session runs on your computer, against your real files, tools and MCP servers, and the phone is the window into it. There are three ways to start it:

• claude --remote-control <name> starts an interactive session you can type into at the terminal and steer from the app, listed under that name.
• /remote-control inside a session you already have open connects it, conversation and all.
• claude remote-control is server mode: the process waits for connections and starts sessions on demand, up to 32 by default. With --spawn worktree, each new session gets its own git worktree, so two never edit the same files. Press the spacebar for a QR code to open on your phone.

In the app, tap Code; a Remote Control session shows a computer icon with a green dot while it is online. Turn on “Push when actions required” or “Push when Claude decides” in /config and Claude Code can notify your phone, or just ask in a prompt: “notify me when the tests finish.” The computer makes outbound connections only and opens no port; while it is connected, the transcript is stored on Anthropic’s servers.

Two conditions come with it. It needs the full sign-in, claude auth login: a setup-token can only make model requests and can’t start Remote Control, and API keys aren’t accepted. And in Anthropic’s words, “your computer has to stay on and the claude process has to keep running.” A laptop that sleeps reconnects when it wakes; nothing ran while it slept. On a remote machine, the docs say to start the session inside tmux or screen so it survives your SSH connection dropping, and server mode gives up and exits after about ten minutes without a network connection.

On a computer that stays on

That last condition is why developers who use Remote Control every day move the sessions off the laptop and onto a machine that is never asleep: a desktop left on, a server, or a cloud computer of their own. On a Linux machine you reach over SSH, server mode inside tmux is the whole setup:

cd ~/code/app
tmux new -s phone
claude remote-control --name app --spawn worktree

The first run asks you to confirm Remote Control and to trust the folder; answer, then detach with Ctrl-b d and the server keeps running. Your phone then sees a session named app, and any new session you start from it gets a worktree of its own. Several sessions at once, with one of them coordinating the rest, is an orchestrator, which needs a little more room than one session. The machine holds your code and your Claude sign-in, so it should be one only you can log in to, and not one open to the internet. Everpod’s developer pod is a machine of this kind: a permanent cloud computer (VPS), set up and secured for you, with your pick of Claude Code, Codex, OpenCode, Pi, Hermes and OpenClaw installed, reached only through your own private network. Codex has its own phone route, with different conditions; see Codex remote control.

Projects: an orchestrator you talk to from the app

Projects, in public beta on Pro and Max, is one ongoing conversation that starts a thread for each task you give it. Its cloud threads keep going after you close your laptop; a thread on your own computer runs through claude remote-control and “only while that computer is awake with Remote Control turned on.” Code has to be on GitHub.

Dispatch and channels

Dispatch is a conversation in the Claude desktop app’s Cowork tab that you message from your phone; it can start a Code session on that computer to handle the task. It needs a Pro or Max plan and the desktop app running.

Channels skip the Claude app: you message a Telegram or Discord bot (or, on a Mac, plain iMessage) and the message lands in the Claude Code session you have open, with the answer coming back in the same chat. Setup is a plugin, a bot token and a pairing code with an allowlist, then restarting Claude with the channel flag:

/plugin install telegram@claude-plugins-official
claude --channels plugin:telegram@claude-plugins-official

Permission prompts can be relayed to the chat so you can approve or deny remotely. The docs’ own condition: “Events only arrive while the session is open, so for an always-on setup you run Claude in a background process or persistent terminal.” If an agent that is always there in your chat app appeals more than a coding tool does, that is the shape OpenClaw was built around.

Plain SSH, and third-party apps

Before any of this existed, people drove Claude Code from phones with an SSH client, and that still works: an always-on machine, the agent in tmux, and an app like Termius or Blink on the phone, which the SSH-from-your-iPhone guide covers. There are third-party control apps too: T3 Code (MIT-licensed, from the t3.gg team) puts Claude Code, Codex and others behind one free app for iOS, Android, web and desktop, driving the agents installed on your computer. Third-party means the app and its connection are theirs, so weigh the trust accordingly.