Claude Code on iPad: What Actually Works
Short answer: yes, you can. Longer answer: there are three genuinely different paths, and which one you pick changes the experience more than the word "yes" suggests.
Why this needs an answer in the first place
iPadOS doesn't ship a terminal, and Claude Code is a command-line TUI. Getting it onto an iPad screen always means going through something else first — an app, an SSH client, or a browser tab talking to a machine that's actually running the terminal.
Three paths
1. The official Claude app
On an iPad, you install the same iOS app you'd put on an iPhone. (Source: official docs) From there you're using either Cloud sessions or Remote Control, same as on a phone. It's the lowest-friction option if you already have an Anthropic account at the right plan tier — no separate SSH setup, no server address to remember.
2. An SSH client + tmux
iPad SSH clients like Blink or Termius connect straight to a dev machine you already have running claude inside tmux. This gets you the real terminal experience — actual keybindings, actual scrollback, nothing translated through a mobile UI layer.
3. Remote desktop / VNC into a real computer
You're not running Claude Code "on" the iPad at all here — you're looking at a full desktop over VNC or similar and Claude Code happens to be one of the windows on it. It works, but it's the heaviest option: more latency, more data, and a full computer's UI squeezed onto a tablet screen that wasn't the target for it.
Where each one runs into a wall
| Path | Where it falls short |
|---|---|
| Official app | Runs the same layout as the iPhone version; Remote Control still requires your own computer to stay powered on and running Claude Code or Claude Desktop — if the machine's off, you need Cloud sessions instead (Source: official docs) |
| SSH client | You manage session persistence yourself (see our piece on surviving SSH disconnects), there's no built-in push notification, and pure keyboard-driven workflows — especially modifier keys like Ctrl and Esc — aren't naturally comfortable on a touchscreen without an external keyboard or a client that adds its own function row |
| VNC / remote desktop | Higher latency, higher data usage, and you're operating an entire computer's interface rather than something purpose-built for a touch display |
One thing worth being direct about, because it's the detail this whole question usually hinges on: none of the above — including the option below — turns an iPad into a purpose-built tablet IDE. What follows is a mobile-shaped tool that also opens on a tablet, not a tablet-native layout.
Where PocketShell fits
PocketShell is a PWA — pure web, no App Store install. Open Safari on the iPad, point it at your own server, and it's there; "add to home screen" gets you a standalone-feeling icon without going through an app review process. Because touch input on a bare terminal is the actual pain point, it ships a full custom on-screen keyboard layer for exactly the keys a physical iPad keyboard-less session is missing — Ctrl, Alt, Esc, arrow keys, and a function-key row — plus a separate IME buffer tab so a system input method (useful for CJK text or emoji) can send a whole composed string at once instead of fighting the terminal character by character.
What it doesn't do, to be direct about the same caveat from above: it isn't a tablet-optimized layout. The project's own scope is explicitly mobile-phone viewports first (roughly 390×844-class), and opening it in an iPad browser gives you that same phone-width interface rather than something that spreads out to use the extra screen the way a native iPad app would. If what you're after is specifically a large-screen, tablet-native coding experience, none of the four options here — including this one — is built for that.