Quick answer
Claude Code still runs on your computer, but your iPhone can monitor it in real time. Using Claude Code hooks wired to an app like Agentfy, you get push notifications, lock-screen Live Activities, and Dynamic Island status the moment an agent finishes, errors, or hits a permission prompt — all without sitting at your desk.
Claude Code lives in your terminal. Your terminal lives on your laptop. So the second you walk away — to grab coffee, to a meeting, to the couch — you lose all visibility into what your agents are doing. That gap is the whole problem, and your iPhone is the obvious place to close it.
This guide covers everything you can actually do with Claude Code from an iPhone today: monitor running agents, get push notifications when they need you, see live status on your lock screen, and keep tabs on several sessions at once. No remote desktop, no jailbreak, no babysitting.
What "Claude Code on iPhone" actually means
Claude Code itself runs on your dev machine — that doesn't change. What you move to your phone is the part that matters when you step away: awareness and control of state. There are three distinct things people mean by this:
- Monitoring — knowing what each agent is doing right now, and getting pinged when one finishes or needs a decision.
- Notifications — a push the instant an agent hits a permission prompt, completes a task, or errors out.
- Full remote control — actually typing back into a session from your phone. Powerful, but heavier to set up and riskier.
For most developers, the first two cover 95% of the pain. You rarely need to drive from your phone — you need to know when to walk back to your desk. The mental model that helps: your Mac stays the cockpit, and your phone becomes the dashboard light that tells you when to look up.
There is no native iOS build of Claude Code, and you don't need one. The goal isn't to recreate a terminal on a touchscreen — it's to push the agent's state to a device you already carry so you stop tabbing back to your laptop every few minutes.
The fastest setup: hooks + a monitor app
Claude Code ships with hooks — events it fires when things happen (a session starts, stops, asks for permission, or sends a notification). Point those hooks at a phone app and you get mobile visibility in about a minute.
- Install Agentfy and grab your token from the app.
- Install the open-source Agentfy plugin — it registers the hooks for you. (On older Claude Code, paste the app's setup prompt to edit
~/.claude/settings.json, or add the hooks by hand.) - Run an agent. Watch the status appear on your phone.
The plugin install is two commands, then a reload:
claude plugin marketplace add ibrillantes/claude-code-plugin
claude plugin install agentfy@agentfy --config api_token=<YOUR_TOKEN>
# then, inside Claude Code:
/reload-plugins
The plugin collects your token in a config dialog and stores it in your OS keychain — not a plaintext file. Because it's open source, you can read exactly which events it sends.
That's the entire flow. Because it rides on hooks rather than a wrapper, there's no daemon to keep alive, no tmux to babysit, and nothing to reattach after a reboot — Claude Code fires the events itself. The relevant hook events look like this:
{
"hooks": {
"Notification": [{ "hooks": [{ "type": "command", "command": "curl -s -X POST <your-webhook-url> ..." }] }],
"Stop": [{ "hooks": [{ "type": "command", "command": "curl -s -X POST <your-webhook-url> ..." }] }]
}
}
You don't have to write that yourself — the plugin (or the setup prompt) does it — but it's worth knowing exactly what's being wired up.
Free covers 25 notifications a month, which is plenty to try it on a real project. Pro unlocks unlimited notifications plus Live Activities, which is the feature most people actually upgrade for.
Full walkthrough here: How to get Claude Code notifications on your iPhone. If pushes set up but never arrive, the usual culprits are covered in Claude Code notifications not working.
Live Activities and the Dynamic Island
This is the part that feels like magic. iOS 18 Live Activities let an app show persistent, updating status on your lock screen and in the Dynamic Island — no unlocking, no opening anything. For Claude Code, that means a glanceable readout like 2 cooking · 1 needs you sitting on your lock screen while you do something else.
When an agent flips to the "your move" state, the island updates instantly. You glance, you see it's a npm test permission, and you know to head back to your desk — you never broke focus chasing a terminal that wasn't doing anything yet. Deep dive: Monitor Claude Code with Live Activities and Dynamic Island.
Stop babysitting your terminal
Agentfy pushes Claude Code status to your iPhone — Live Activities, Dynamic Island, and instant alerts.
Running several agents at once
The phone really earns its place when you're running more than one agent. Three terminals across two projects is impossible to track by tabbing between windows — but trivial when each one reports to a single consolidated view on your phone. Instead of mentally juggling which window was waiting on what, you get one feed that tells you which session needs attention and which are still working.
If that's your workflow, read how to run multiple Claude Code agents in parallel next. It pairs naturally with the broader vibe coding loop, where you fan out work to several agents and only step in at decision points.
What about full remote control?
If you genuinely need to type back into a session from your phone, your options are a remote terminal (SSH via an app like Blink or Termius into a tmux session) or a tool built specifically for remote Claude Code control. These work, but they come with tradeoffs: you're exposing a shell, typing commands on a touch keyboard, and taking on the security of whatever tunnel you set up.
Here's how the two approaches compare:
| Approach | What you can do | Setup effort | Risk profile | | --- | --- | --- | --- | | Hooks + monitor (Agentfy) | See status, get notified, glance at Live Activities | Under a minute, one plugin install | Low — only session state leaves your machine | | SSH terminal (Blink, Termius + tmux) | Full control: type commands back | Higher — keys, tunnel, multiplexer | Higher — you expose a real shell |
Agentfy is monitoring-only today — you can see when an agent needs you, but you can't approve a permission or reply from the phone yet. If typing back is a hard requirement right now, an SSH terminal is your route. For most people, knowing when to walk back to the desk is enough.
We compared the full landscape — notification-only tools through full remote terminals — in the best apps to monitor Claude Code from your phone.
Is any of this private?
With a hooks-based monitor, the only thing leaving your machine is session state — a project name, a status, and the permission context (for example, Bash: npm test). Your code and file contents stay local; if you turn on the optional prompt preview, it's truncated to the first 30 characters on your device before anything is sent. A remote terminal is different: that's a real shell, so treat it with the same care you'd give any SSH endpoint.
That difference is the main reason most developers land on the monitoring approach. You get the awareness without widening your attack surface, and you can see exactly what each hook sends — the plugin is open source, and on the manual path it's all in your own settings file.
The short version
You don't need to run Claude Code on your iPhone. You need your iPhone to tell you what Claude Code is doing on your Mac. Hooks plus a monitor app gets you there in a minute, and Live Activities make it effortless to glance at.
Agentfy is free to start and does exactly this — agent status on your lock screen, push the moment one needs you. It requires iOS 18 or later, since Live Activities and the Dynamic Island are what make the experience feel seamless.