swift-architecture — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited swift-architecture (Agent Skill) and scored it 45/100 (orange). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 2 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 2 flagged
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
*.sig, SIGNATURES) outside the documentation.Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Choose the smallest architecture that makes state ownership, dependencies, side effects, and tests explicit. Default new SwiftUI features to MV; escalate only for observed complexity.
This skill owns pattern selection, module boundaries, dependency direction, migration strategy, and architecture-level test seams. Route SwiftUI property-wrapper wiring and view composition to swiftui-patterns, navigation APIs and route models to swiftui-navigation, isolation diagnostics to swift-concurrency, and test syntax/fixtures to swift-testing.
| Pattern | Choose when | Main cost |
|---|---|---|
| MV | SwiftUI feature has straightforward state and orchestration | Logic can drift into large views without decomposition |
| MVVM | Presentation logic needs an independently testable adapter | Extra layer can become a forwarding shell |
| MVI | A feature is best modeled as explicit state + intents + reducer/effects | Boilerplate and centralized transition design |
| TCA | Many composable features need deterministic effects, dependencies, and testing | Framework learning and architectural commitment |
| Clean Architecture | Large product needs strict dependency direction across domain/data/UI | Protocol and mapping overhead |
| Coordinator | UIKit or hybrid navigation needs a separate flow owner | Another lifecycle and routing owner |
| VIPER | Maintaining an existing UIKit module with established VIPER boundaries | Very high ceremony; poor default for new SwiftUI work |
Use Coordinator alongside another state pattern when navigation complexity is the pressure; it is not a replacement for domain/state architecture.
Keep views as state expressions and put business operations in observable models and injected services:
@MainActor
@Observable
final class TripStore {
private let client: TripClient
var trips: [Trip] = []
var error: Error?
init(client: TripClient) { self.client = client }
func load() async {
do { trips = try await client.fetchTrips() }
catch { self.error = error }
}
}
struct TripList: View {
@State private var store: TripStore
init(client: TripClient) {
_store = State(initialValue: TripStore(client: client))
}
var body: some View {
List(store.trips) { Text($0.name) }
.task { await store.load() }
}
}Load Architecture Pattern Recipes for MVVM, MVI, TCA, Clean Architecture, Coordinator, and VIPER structure.
Do not escalate merely because a view is long. First extract subviews, services, and focused observable models.
Migrate one feature boundary at a time:
For ObservableObject to Observation, preserve the same owner and mutation isolation before replacing wrappers. For MVVM to MV, delete forwarding view-model members only after views bind to the same model/service behavior. For TCA adoption, wrap one feature's state/actions/effects and migrate dependencies incrementally.
| Mistake | Fix |
|---|---|
| Pattern chosen by popularity | Tie it to an observed feature pressure. |
| View model only forwards properties | Remove it and use MV. |
| One object owns navigation, networking, formatting, persistence, and UI state | Split by responsibility and dependency direction. |
| TCA or Clean Architecture applied to trivial screens | Start with MV and preserve an escalation seam. |
| Coordinator used as a state architecture | Keep it focused on route/lifecycle ownership. |
| Multiple patterns mixed inside one feature | Define one local state/effect model and migrate at feature boundaries. |
| Big-bang migration | Move one tested vertical slice and rerun the same proof matrix. |
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.