App Intent Driven Development

App Intent–First Driven Development

Design features as App Intents first, then reuse those intents across Shortcuts, widgets, and SwiftUI views so automation and UI stay in lockstep.

Core Ideas

• Entities first: model the data users act on (events, categories, records) as AppEntity so intents, widgets, and the app share one source of truth.
• Intent-first feature: build the App Intent + entity query before UI; SwiftUI screens call those intents instead of duplicating service code.
• Single action, single intent: keep intents focused; avoid mega-intents that are hard to compose in Shortcuts.
• Predictable UI: supply DisplayRepresentation, typeDisplayRepresentation, and icons so Siri/Shortcuts can render rich cards without opening the app.
• Fast queries: EntityQuery must be quick and cancellable; avoid blocking the main actor.
• Reuse business logic: intents call the same services your views use; do not fork logic inside the intent.

Minimal Entity Blueprint

import AppIntents

struct TaskEntity: AppEntity, Identifiable {
    static let typeDisplayRepresentation = TypeDisplayRepresentation(name: "Task")
    static let defaultQuery = TaskQuery()

    let id: UUID
    let title: String
    let isComplete: Bool

    var displayRepresentation: DisplayRepresentation {
        DisplayRepresentation(
            title: title,
            subtitle: isComplete ? "Completed" : "Open",
            image: .init(systemName: isComplete ? "checkmark.circle.fill" : "circle")
        )
    }
}

struct TaskQuery: EntityQuery {
    func entities(for identifiers: [UUID]) async throws -> [TaskEntity] {
        try await TaskStore.shared.fetch(ids: identifiers) // fast path
    }

    func suggestedEntities() async throws -> [TaskEntity] {
        try await TaskStore.shared.fetchRecent()
    }
}

Key points: stable identifier, meaningful representation, and fast queries that avoid launching heavy app flows.

Intent Pattern

import AppIntents

struct CompleteTaskIntent: AppIntent {
    static let title: LocalizedStringResource = "Complete Task"
    static let description = IntentDescription("Marks a task as done and returns the updated item.")

    @Parameter(title: "Task", requestValueDialog: "Which task should I complete?")
    var task: TaskEntity

    // Used so we can call the intent from SwiftUI using .perform()
    init(task: TaskEntity) { self.task = task }

    @MainActor
    func perform() async throws -> some IntentResult & ReturnsValue {
        let updated = try await TaskStore.shared.complete(task.id)
        return .result(value: updated)
    }

    static var parameterSummary: some ParameterSummary {
        Summary("Complete \(\.$task)")
    }
}

• Parameters: keep them few; provide requestValueDialog to make Siri prompts natural.
• Results: return entities when possible; system surfaces render them nicely.
• Isolation: mark with @MainActor only if you must touch UI-bound objects; otherwise keep work off the main actor.

Reusing Intents in SwiftUI

• Prefer calling intents from UI so automation and in-app flows share one path.
• Use AppIntentButton to invoke intents directly from views.
• Translate entity selections into view state so widgets/Shortcuts and in-app pickers present the same objects.

import AppIntents
import SwiftUI

struct EventRow: View {
    let event: EventEntity

    var body: some View {
        HStack {
            Text(event.name)
            Spacer()
            AppIntentButton(intent: UndoLastEventOccuranceIntent(event: event)) {
                Label("Undo", systemImage: "arrow.uturn.backward")
            }
        }
    }
}

• For more control, invoke intents imperatively with perform (e.g., to show progress or handle errors):

import AppIntents
import SwiftUI

struct EventRow: View {
    @Environment(\.intentExecutor) private var executor
    @State private var isWorking = false
    @State private var error: Error?

    let event: EventEntity

    var body: some View {
        HStack {
            Text(event.name)
            Spacer()
            Button {
                Task {
                    isWorking = true
                    defer { isWorking = false }
                    do {
                        try await executor.perform(UndoLastEventOccuranceIntent(event: event))
                    } catch {
                        self.error = error
                    }
                }
            } label: {
                if isWorking {
                    ProgressView()
                } else {
                    Label("Undo", systemImage: "arrow.uturn.backward")
                }
            }
        }
        .alert("Undo failed", isPresented: .init(
            get: { error != nil },
            set: { if !$0 { error = nil } }
        )) {
            Button("OK", role: .cancel) { error = nil }
        } message: {
            Text(error?.localizedDescription ?? "Unknown error")
        }
    }
}

• Keep the intent signature identical between Shortcuts and SwiftUI usage.
• Avoid reimplementing service calls in views; route through the intent to keep analytics, validation, and side effects consistent.

Development Flow

1. Model the domain type as AppEntity with DisplayRepresentation and EntityQuery.
2. Implement a focused AppIntent that calls shared services; avoid duplicate data access layers inside the intent.
3. Add previews in Shortcuts or the App Intents preview panel; ensure suggested entities show immediately.
4. Expose the same entity in widgets/Live Activities to keep automation and UI consistent.
5. Localize strings early (LocalizedStringResource) to keep Siri responses natural in all supported languages.

Quick Checklist

• [ ] Entity has stable id, typeDisplayRepresentation, and rich displayRepresentation.
• [ ] Queries are fast, cancellable, and return suggestions without opening the app.
• [ ] Intent reuses shared domain services; no duplicated business logic.
• [ ] Parameters are minimal and well phrased with requestValueDialog.
• [ ] Results return entities when possible for better system rendering.
• [ ] Strings are localized; tests cover queries and perform paths.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: mintuz
• Source: mintuz/skills
• License: MIT

Install and usage instructions live in the source repository linked above.