Introducing CFA AI Toolkit for iOS v0.3.0
AI can generate a SwiftUI screen in seconds. Building an iOS app you can understand, extend, and maintain takes more than a fast first draft.
Today we’re releasing CFA AI Toolkit for iOS v0.3.0, a free, open-source toolkit from 3 Days of Swift Concurrency. It gives AI coding tools a consistent way to create and work on iOS apps using Cooperative Feature Architecture (CFA), SwiftUI, and Swift Concurrency.
This release brings the downloadable toolkit up to date with the project we’ve been building. It includes eight AI skills, covering app creation, feature work, architecture adoption, GCD migration, architecture review, codebase tidying, unit tests, and Swift code quality. It also includes a native CFA App Xcode template and the cfa command for inspecting a project, creating a local architecture dashboard, and checking an installation.
The idea is straightforward: when you ask an AI to add a feature, it should have a clear place to put the code and clear responsibilities to preserve. In a CFA app, each feature has a named home. The app’s structure is visible in Xcode, rather than hidden in a long prompt or scattered across unrelated folders.
The toolkit is useful whether you’re starting a new app or looking after one that already exists. You can create a project from the Xcode template, ask your AI coding tool to add a CFA feature, or use the inspector to understand an existing codebase before changing it. CFA does not automatically rewrite an existing app, and its reports are evidence for review—not a certificate that an app is race-free or ready to ship.
We’ve also made the release easier to try. The downloadable plugin contains the eight skills and installer. The Xcode template has its own installer, so you can choose whether to add it to Xcode. The project is open source under the MIT licence.
The best way to judge CFA is to use it on an app. Give your AI coding tool a real feature request. Look at the result in Xcode. Ask it to make a second change. Can you still find the code, understand who owns the work, and test what matters? That is the standard we’re building toward.
Comments