Claude Code Desktop Adds an iOS Simulator Pane: Watch Claude Test Your App Live

Claude Code Desktop now shows your iOS app running live in a dedicated iOS Simulator pane next to the conversation. When Claude builds, installs, launches, or checks your app in a simulator, the pane opens automatically, streams the device screen, and Claude taps through the app and reads the screen to verify its own changes while you watch, or you can tap through it yourself (Anthropic docs). The pane is in public beta on macOS for Pro, Max, and Team plans (not Enterprise).

Key Takeaways

• The pane opens automatically when Claude runs your app in a simulator, with no command or setting needed. Ask "build the app and run it in the simulator to check the onboarding flow" and it appears (Anthropic docs).
• It drives the simulator directly, without computer use, so it never takes over your screen and does not need macOS Accessibility or Screen Recording permissions.
• The pane is interactive: you and Claude share the same device. Tap, swipe, rotate, press hardware buttons, and save screenshots (Cmd+S) or recordings (Cmd+R) to your Desktop.
• Requirements: Claude Desktop v1.24012.0+, macOS, and Xcode with the iOS platform installed. Local sessions only; cloud and SSH sessions cannot reach your Mac's simulators.
• Each session owns its devices (up to 4 panes per session), parallel sessions do not share them, and Claude shuts down simulators it booted 10 minutes after you detach, on archive, or on quit.
• Consent is per device, once: the first time Claude uses a simulator you approve it, and Claude's screenshots of the device go to Anthropic under your conversation retention settings, so do not sign in to real accounts on a device Claude uses.

How It Works

Open your iOS project as the project folder in the Code tab, then phrase the task around running or verifying: Claude builds the app, launches it in a simulator, and the pane opens streaming the device. Claude then installs, taps through, and reads the screen to verify its changes. Name a device ("run it on the iPhone SE simulator") and Claude targets it. You can also attach a simulator manually from the Views menu once the session has touched Swift files.

Two actions still follow your session's permission mode rather than the one-time device consent: opening a URL on the device (a URL can carry data off the device) and building the app (xcodebuild runs your project's build scripts on your Mac). Organizations can disable the tools entirely via the disableMobileSimulatorTools managed setting.

The Verification Loop Is the Point

The pane closes the gap that has made mobile agent work awkward: the agent could edit Swift all day, but seeing whether the UI actually looks right meant screen-control hacks or you playing screenshot courier. Now the check-your-own-work loop (build, launch, tap, read the screen, fix) runs in one pane with no screen takeover, and you can grab the device mid-session to navigate somewhere and ask Claude "does this look right?"

The competitive context: OpenAI's answer to iOS development is the Codex build-iOS-apps plugin, and Codex reaches simulators through computer use, which controls your actual screen. Anthropic's pane sidesteps screen control entirely, a meaningfully calmer experience if you work while the agent tests.

How to Use This Update

If you build iOS apps on a Pro, Max, or Team plan, update Claude Desktop past v1.24012.0, open an app project, and ask Claude to run it in the simulator and tap through your most fragile flow. Watch the first run rather than tabbing away; the "Claude is using this device" badge shows you exactly when it is driving versus when the pane is yours. And keep test accounts only on simulator devices Claude uses, since its device screenshots are sent to Anthropic under your retention settings.