Claude Code for iOS development

The series on building with AI was about working with the model in general: the rules, the memory, who decides, and what it reports versus what actually happened. This one is the iOS part, the things that only show up once the model has to build, run and ship an app: the simulator, the tests, and everything the App Store wants on top of the app itself.

It comes from 3 apps: Reckon, a decision journal, and Continuum, for keeping track of the people you manage, which are both out, and Digest, a food journal I'm working on. Claude wrote the code for all 3, mostly through Claude Code, while I decided what it should be and checked what came back.

My setup is the Claude Code desktop app, with its iOS Simulator pane and xcodebuild and xcrun simctl running through Bash, a handful of Swift skills and the rules and hooks that keep it working the way I want. I used Xcode's own agent first, and why I moved is part of the first post.

Some of this is specific to my apps and my setup, and Claude Code changes fast enough that a few details will probably age. The problems behind them change a lot slower, though: a tool that fails and points you at the wrong cause, a test that only checks the code against itself, or a result nobody can verify. So most posts start from the problem and use my apps as the example.

What the series covers

• Claude Code and the iOS Simulator: what breaks — Claude Code drives the simulator from the desktop app with no MCP server; the sandbox, misleading xcodebuild errors and dead captures, and the second check that catches each one.
• iOS testing with Claude Code — tests written from the code end up defending its bugs; where the expected values should come from instead, and why breaking the code is how I check a test.
• Testing code when you don’t know the right answer — plant the answer in data you invent, plant nothing as well, and write down what passing means before the first run.
• Automating App Store screenshots with Claude Code — the app draws its own screenshots from its real components, and Claude Code runs the export.
• App Store preview videos with Claude Code — the app’s screens redrawn in code instead of recorded.
• A/B testing App Store screenshots on a small app — test promises that differ a lot, decide by judgment when the traffic can’t, and upload the treatments with a script.

If you're building iOS apps with Claude Code too, I'd love to hear what broke for you @roland.leth.ro or @rolandleth.