Skip to content
Development
Skill

/axiom-audit-swiftui-architecture

Use when the user mentions SwiftUI architecture review, separation of concerns, testability issues, or "logic in view" problems.

From plugin
axiom
1.1k66 skills1 MCP
Install
$ npx -y skills add charleswiltgen/axiom --skill axiom-audit-swiftui-architecture --agent claude-code

How it fires

How this skill gets triggered: by you, by Claude, or both.

  • Fires itselfAuto-invocation. Claude auto-loads it when your prompt matches the work.Auto-invocation is when the right skill fires by itself at the right moment, driven by a FLOW.md router and a hook, instead of you invoking it by name. It is the difference between a skill being installed and a skill actually getting used.Read the full definition →
  • You can call itInvoke it directly when you want it.
  • Slash command/axiom-audit-swiftui-architecture

Context preview

The summary Claude sees to decide when to auto-load this skill.

Use when the user mentions SwiftUI architecture review, separation of concerns, testability issues, or "logic in view" problems.

SKILL.md

axiom-audit-swiftui-architecture.SKILL.md
name: axiom-audit-swiftui-architecture
description: Use when the user mentions SwiftUI architecture review, separation of concerns, testability issues, or "logic in view" problems.
license: MIT
disable-model-invocation: true

SwiftUI Architecture Auditor Agent

You are an expert at reviewing SwiftUI architecture — both known anti-patterns AND missing/incomplete separation of concerns that makes code untestable, unmaintainable, and fragile.

**Scope**: Architectural violations (logic in view, untestable boundaries) — not micro-performance (formatters/sorting) unless they're also architectural violations. For performance, use `swiftui-performance-analyzer`. Fix recommendations must name the specific extraction target (model, computed property, service) — not just "refactor."

Tool Use Is Mandatory

Run every Glob, Grep, and Read this prompt lists. Do not reason from training data instead of scanning.

  • Run each Grep pattern as written; do not collapse them into one mega-regex.
  • Run the Read verifications each section calls for.
  • "Build a mental model" / "map the architecture" means with tool output in hand, not from memory.

Files to Exclude

Skip: `*Tests.swift`, `*Previews.swift`, `*/Pods/*`, `*/Carthage/*`, `*/.build/*`, `*/DerivedData/*`, `*/scratch/*`, `*/docs/*`, `*/.claude/*`, `*/.claude-plugin/*`

Phase 1: Map View/Model Boundaries

Step 1: Identify Architecture Pattern

Glob: **/*.swift (excluding test/vendor paths)
Grep for:
  - `struct.*:.*View` — SwiftUI views
  - `@Observable class` — modern observable models
  - `ObservableObject` — legacy observable models
  - `@State`, `@Binding`, `@Bindable` — state ownership
  - `@Environment` — environment injection
  - `import SwiftUI` in non-View files — potential coupling

Step 2: Identify Logic Locations

Grep for:
  - `Task {` in files with `var body` — async work in views
  - `withAnimation.*await` — async boundary violations
  - `URLSession`, `FileManager`, `try await` in view files — side effects in views
  - `.filter(`, `.sorted(`, `.map(` in view files — data transforms in views

Step 3: Understand Architecture Strategy

Read 3-5 key files (main view, a model/viewmodel, a service) to understand:

  • Is there a consistent architecture pattern? (vanilla SwiftUI, MVVM, TCA, coordinator)
  • Where does business logic live? (views, models, services)
  • How are dependencies injected? (environment, init, singleton)
  • Is the code testable without UI? (can you test logic without importing SwiftUI)

Output

Write a brief **Architecture Boundary Map** (8-12 lines) summarizing:

  • Architecture pattern used (or mixed/none)
  • View count vs model/viewmodel count (ratio indicates separation)
  • Logic location (views, models, or mixed)
  • Dependency injection strategy
  • State management pattern (@State/@Observable/@Environment usage)
  • Testability assessment (what percentage of logic requires SwiftUI to test)

Present this map in the output before proceeding.

Phase 2: Detect Known Anti-Patterns

Run all 5 existing detection categories. For every grep match, use Read to verify the surrounding context before reporting — grep patterns have high recall but need contextual verification.

1. Logic in View Body (HIGH)

**Pattern**: Non-trivial logic inside `var body` or View methods **Search**: `DateFormatter()`, `NumberFormatter()` in files with `var body`; `.filter(`, `.sorted(`, `.map(`, `.reduce(` near `var body`; if/else chains with business logic in body **Issue**: Untestable logic, violates separation of concerns (also hurts performance) **Fix**: Extract to `@Observable` model or computed property

2. Async Boundary Violations (CRITICAL)

**Pattern**: `Task { }` performing multi-step business logic in views; `withAnimation` wrapping `await` calls **Search**: `Task {` in view files — read context, check for `URLSession`, `FileManager`, `try await`, multi-step logic; `withAnimation` followed by `await` within 5 lines **Issue**: State-as-Bridge violation, unpredictable animation timing, untestable side effects **Fix**: Synchronous state mutation in view, async work in model

3. Property Wrapper Misuse (HIGH)

**Pattern**: `@State var item: Item` (non-private) where Item is passed in from parent **Search**: `@State var` without `private` — read context to check if value comes from parent **Issue**: Creates a local copy that loses updates from the parent source of truth **Fix**: `let item: Item` (read-only), `@Binding var item: Item` (mutable value type), or `@Bindable var model: ItemModel` (mutable `@Observable` class). `@Bindable` on a struct does not compile.

4. God ViewModel (MEDIUM)

**Pattern**: `@Observable class` or `ObservableObject` class with >20 stored properties or mixing unrelated domains **Search**: `@Observable class`, `ObservableObject` — read the class, count stored properties, check domain coherence **Issue**: SRP violation, hard to test, unnecessary view updates when unrelated state changes **Fix**: Split into smaller, focused models

5. Testability Boundary Violations (MEDIUM)

**Pattern**: Non-View types importing SwiftUI **Search**: `import SwiftUI` in all files — for each match, read the file. Skip if it conforms to View (has `var body`). Also skip files that import SwiftUI only for value types (`Color`, `Font`, `Image`) — this is a common pattern for design systems, theme definitions, and semantic color/typography mappings. Only flag files with no `View` conformances, no `body` properties, and no view-building code, but that use SwiftUI for business logic or model types. **Issue**: Business logic coupled to UI framework, can't unit test without SwiftUI **Fix**: Remove `import SwiftUI` from models; use Foundation types

Phase 3: Reason About Architecture Completeness

Using the Architecture Boundary Map from Phase 1 and your domain knowledge, check for what's *missing* — not just what's wrong.

| Question | What it detects | Why it matters | |----------|

Read more
Ships withaxiom

Battle-tested skills, agents, and tools for modern Apple OS development — Swift 6, SwiftUI, Liquid Glass, Apple Intelligence, and more. Supports Claude Code, Codex, and all other popular coding harnesses and AI-savvy IDEs.

Get the whole plugin