LLMs often produce SwiftUI code that looks correct but uses soft-deprecated APIs, breaks accessibility, or introduces subtle performance problems. This project packages practical, battle-tested rules so coding agents can reliably spot and fix those issues during code generation and review.
What Sets It Apart
- Focused on LLM failure modes: checks target problems LLMs commonly introduce (invisible controls to VoiceOver, deprecated modifiers, expensive work in view bodies), so fixes are high-impact for AI-generated code.
- Agent-native format: provided as an Agent Skills package (SwiftUI Pro) so it plugs into Claude Code, Codex, Gemini, Cursor, and similar assistants — enabling automated reviews and in-context guidance rather than just a static checklist.
- Practical, opinionated guidance: recommends modern Swift/SwiftUI idioms (avoid certain Text concatenations, prefer newer formatting APIs, use @Observable where helpful) and pairs each check with actionable replacements, not just warnings.
- Complements tooling: designed to augment, not replace, linters or human review — it focuses on stylistic, accessibility, and agent-specific pitfalls that static analyzers may miss.
Who It's For and Trade-offs
Great fit if you write SwiftUI and rely on AI coding assistants to generate or review code — it reduces common post-processing and bugfix cycles. It’s also useful for teams adopting Swift 6+ and iOS 26+ conventions who want consistent, agent-driven reviews. Look elsewhere if you need a runtime library, CI-only static analyzer, or a general-purpose Swift linter: this is guidance for agents and reviewers, not a compile-time enforcement tool. Expect some integration work with your chosen agent and modest token cost when running comprehensive checks.
