Check on Claude Code from your phone
Coding agents like Claude Code work for minutes or hours at a time, and stop to ask before doing anything risky. Pocketmux lets you leave your desk and still see what the agent is doing, answer its questions, or stop it, from your iPhone or Android phone. (Pocketmux itself was built with Claude Code.)
How it works
- 1. Start a tmux session with pmux:
pmux new -s agent - 2. Run your agent in it:
claude - 3. Step away. The agent keeps running in tmux on your computer, whether or not your phone is connected.
- 4. Open Pocketmux. Your
agentsession is listed by name; pick the pane and you are looking at the live terminal. The pane list shows what is running in each pane (for Claude Code, that is the version number the CLI reports).
Answer prompts and steer from your phone
Claude Code asks for permission before it edits files or runs commands, and offers choices you pick with the arrow keys. The bar above your phone's keyboard adds the keys phones don't have — Esc, Tab, Ctrl, Shift, Alt, the arrow keys, and ⌃C — so you can approve a tool call, choose an option, interrupt with Esc, or type a follow-up prompt.
Pocketmux passes the raw terminal stream through untouched, so the agent's interface looks the same on your phone as it does on your desktop.
Built for checking in on the go
Phones switch networks and lose signal. If the connection drops, the app and the agent re-establish it and put you back in the same pane. The agent itself never notices: it is running in tmux on your computer, not on your phone.
Keep your agent's machine closed
An agent session has your code, your credentials, and your tokens. Reaching it over SSH means opening a port or running a VPN, and keeping a key on your phone. With Pocketmux there are no open ports: your phone connects directly to your computer over an end-to-end encrypted channel, authenticated with a key bound to your phone, and the signaling server never sees terminal content. The details are in our security model.
Works with any terminal agent
Nothing here is specific to Claude Code. Any agent that runs in a terminal works the same way, such as OpenAI Codex CLI, Gemini CLI, or Aider, along with long builds, test runs, and anything else you'd leave running in tmux.
What it doesn't do (yet)
Pocketmux has no push notifications. It won't alert you when an agent finishes or is waiting for approval; you open the app and check in when it suits you. For a side-by-side look at the alternatives, see Pocketmux vs Termius, Blink Shell, and mosh.
Get started
Install the free, open-source pmux CLI on macOS, Linux, or Windows (via WSL), pair your phone once, and start your next agent session with pmux instead of tmux:
brew install shiftinbits/tap/pmux Setup guides are in the docs. The Pocketmux app is available in the United States and Canada.