Show HN: Axiom for Claude Code – Coding skills for iOS devs(charleswiltgen.github.io)
charleswiltgen.github.io
Show HN: Axiom for Claude Code – Coding skills for iOS devs
https://charleswiltgen.github.io/Axiom/
1 comments
Kind of sad how this hasn‘t gotten any traction. I can tell you’ve put a lot of thought into this. Native iOS development with agents is a pain without a proper system to support the LLM. I will give this a spin once I have some spare time. Thank you for sharing this!
I recently christened it v1.0, but I continue to evolve it quickly. It's biased toward modern frameworks and best practices, but I continue to notably broaden coverage every week. Here are the technologies and concepts Axiom currently has deep coverage for:
• Swift & Concurrency — Swift 6 strict concurrency, actors, Sendable, @MainActor, async/await, data race prevention, actor isolation patterns
• SwiftUI — Performance optimization with Instruments, navigation architecture (NavigationStack, NavigationSplitView, deep linking), adaptive layout (ViewThatFits, AnyLayout), debugging view updates and preview crashes, gestures, Liquid Glass
• Data & Sync — SwiftData, SQLiteData (Point-Free), GRDB, Core Data (migrations, thread-confinement, N+1 queries), Codable/JSON, CloudKit (CKSyncEngine, conflict resolution), iCloud Drive, file protection, Realm → SwiftData migration
• Debugging & Performance — Memory leak patterns (timers, observers, closures, delegates), Instruments workflows (Time Profiler, Allocations, Leaks), Xcode build failures, Network.framework diagnostics, Auto Layout conflicts, build optimization
• Design & Accessibility — Human Interface Guidelines, VoiceOver, Dynamic Type, WCAG compliance, color contrast, touch targets, typography (San Francisco, text styles)
• Apple Frameworks — Foundation Models/Apple Intelligence, App Intents (Siri, Shortcuts, Spotlight), StoreKit 2 (subscriptions, transactions), AVFoundation (audio sessions, spatial audio), WidgetKit, Live Activities, TextKit 2, Now Playing
As with all useful AI tools, what makes Axiom perform so much better than foundation models alone is the thoughtful curation and optimization of the knowledge management at the heart of its capabilities. For example, I've spent a bunch of time "encoding" knowledge from WWDC sessions into relevant components, which gives Axiom a unique and useful "POV" that, in many cases, isn't covered in Apple's reference docs.
Any and all feedback/criticism is welcome, and I'll use it to improve Axiom for everyone. If you use Claude Code, I'd appreciate if you could try one or two of the audit commands (https://charleswiltgen.github.io/Axiom/commands/) on your codebase. You can also use `/axiom:ask` to ask Axiom about it's capabilities, what skills and agents would be most useful for a given issue or area of your app, etc.