Relevancekit logo

Relevancekit

CommunityPopular
dpearson2699
relevancekit

Increase widget visibility on Apple Watch using RelevanceKit. Use when providing contextual relevance signals for watchOS widgets, declaring time-based or location-based relevance, combining multiple relevance providers, helping the system surface the right widget at the right time on watchOS 26, or routing mixed RelevanceKit/WidgetKit/HealthKit/MapKit Smart Stack scope.

Overview

Publisherdpearson2699
Repositoryswift-ios-skills
Skill namerelevancekit
Stars
1.2K
Forks
61
Bundled files
2
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 2 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by dpearson2699 on GitHub. Read the source before you install it.

Installation

Install the Relevancekit AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/dpearson2699/swift-ios-skills/tree/main/skills/relevancekit
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/dpearson2699/swift-ios-skills.git /tmp/swift-ios-skills
mkdir -p .claude/skills
cp -r /tmp/swift-ios-skills/skills/relevancekit .claude/skills/relevancekit
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Relevancekit in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Relevancekit on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Relevancekit is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

RelevanceKit

Provide on-device contextual clues that increase a widget's visibility in the Apple Watch Smart Stack. RelevanceKit tells the system when a widget is relevant by time, location, fitness state, sleep schedule, or connected hardware. Targets Swift 6.3 / watchOS 26+.

Beta-sensitive. Re-check Apple documentation before making strong RelevanceKit availability or behavior claims.

See references/relevancekit-patterns.md for complete relevant-widget, timeline provider, grouping, preview, and permission patterns.

Contents

Overview

watchOS uses two mechanisms to determine widget relevance in the Smart Stack:

  1. Timeline provider relevance -- implement relevance() on an existing AppIntentTimelineProvider to attach RelevantContext clues to timeline entries. Available across platforms; only watchOS acts on the data.
  2. Relevant widget -- use RelevanceConfiguration with a RelevanceEntriesProvider to build a widget driven entirely by relevance clues. The system creates individual Smart Stack cards per relevant entry. watchOS 26+ only.

Choose a timeline provider when the widget always has data to show and relevance is supplementary. Choose a relevant widget when the widget should only appear when conditions match, or when multiple cards should appear simultaneously (e.g., several upcoming calendar events).

Key Types

TypeModuleRole
RelevantContextRelevanceKitA contextual clue (date, location, fitness, sleep, hardware)
WidgetRelevanceWidgetKitCollection of relevance attributes for a widget kind
WidgetRelevanceAttributeWidgetKitPairs a widget configuration with a RelevantContext
WidgetRelevanceGroupWidgetKitControls grouping behavior in the Smart Stack
RelevanceConfigurationWidgetKitWidget configuration driven by relevance clues (watchOS 26+)
RelevanceEntriesProviderWidgetKitProvides entries for a relevance-configured widget (watchOS 26+)
RelevanceEntryWidgetKitData needed to render one relevant widget card (watchOS 26+)

RelevanceConfiguration, RelevanceEntriesProvider, and RelevanceEntry are WidgetKit APIs. Keep them in this skill's scope only when they are part of the watchOS relevant-widget workflow that exposes RelevanceKit clues.

Setup

Import

swift
import RelevanceKit
import WidgetKit

Platform Availability

RelevantContext is declared across platforms (iOS 17+, watchOS 10+), but RelevanceKit functionality only takes effect on watchOS. Calling the API on other platforms has no effect. Timeline-provider relevance() is available on iOS 18+, macOS 15+, visionOS 26+, and watchOS 11+ for shared provider code. RelevanceConfiguration, RelevanceEntriesProvider, and RelevanceEntry are watchOS 26+ only.

Permissions

Certain relevance clues require authorization or target setup:

ClueRequired Permission
.location(inferred:)Containing app requests location access; widget extension declares NSWidgetWantsLocation
.location(_:) (CLRegion)Containing app requests location access; widget extension declares NSWidgetWantsLocation
.location(category:)Containing app requests location access; widget extension declares NSWidgetWantsLocation
.fitness(.workoutActive)HealthKit access to HKWorkoutType
.fitness(.activityRingsIncomplete)HealthKit access to appleExerciseTime, appleMoveTime, and appleStandTime
.sleep(_:)HealthKit sleepAnalysis permission
.hardware(headphones:)None
.date(...)None

Add location purpose strings to the containing app's Info.plist, not only the widget extension. In widget code, check CLLocationManager.isAuthorizedForWidgetUpdates before relying on location clues. For fitness and sleep clues, enable HealthKit and request the exact read types in the app and widget extension target that provides relevance.

Relevance Providers

Option 1: Timeline Provider with Relevance

Add a relevance() method to an existing AppIntentTimelineProvider. This approach shares code across iOS and watchOS while adding watchOS Smart Stack intelligence.

swift
struct MyProvider: AppIntentTimelineProvider {
    // ... snapshot, timeline, placeholder ...

    func relevance() async -> WidgetRelevance<MyWidgetIntent> {
        let attributes = events.map { event in
            let context = RelevantContext.date(
                from: event.startDate,
                to: event.endDate
            )
            return WidgetRelevanceAttribute(
                configuration: MyWidgetIntent(event: event),
                context: context
            )
        }
        return WidgetRelevance(attributes)
    }
}

Option 2: RelevanceEntriesProvider (watchOS 26+)

Build a widget that only appears when conditions match. The system calls relevance() to learn when the widget matters, then calls entry() with the matching configuration to get render data.

swift
@available(watchOS 26.0, *)
struct MyRelevanceProvider: RelevanceEntriesProvider {
    func relevance() async -> WidgetRelevance<MyWidgetIntent> {
        let attributes = events.map { event in
            WidgetRelevanceAttribute(
                configuration: MyWidgetIntent(event: event),
                context: RelevantContext.date(event.date, kind: .scheduled)
            )
        }
        return WidgetRelevance(attributes)
    }

    func entry(
        configuration: MyWidgetIntent,
        context: Context
    ) async throws -> MyRelevanceEntry {
        if context.isPreview {
            return .preview
        }
        return MyRelevanceEntry(event: configuration.event)
    }

    func placeholder(context: Context) -> MyRelevanceEntry {
        .placeholder
    }
}

Boundary Routing

When a feature mixes widgets, location, workouts, and Smart Stack relevance, keep RelevanceKit focused on RelevantContext, WidgetRelevanceAttribute, provider relevance(), RelevantIntentManager, relevant-widget handoffs, and permissions for relevance clues. Route timelines, reload budgets, families, rendering, APNs widget pushes, Live Activities, and widget Controls to WidgetKit; HKWorkoutSession, HKLiveWorkoutBuilder, HKWorkoutRoute, queries, activity-ring/sleep data, and authorization UX to HealthKit; and MKLocalSearch, MKLocalSearchCompleter, MKDirections, geocoding, authorization, regions, geofencing, and place data to MapKit/CoreLocation.

Time-Based Relevance

Time clues tell the system a widget matters at or around a specific moment.

Single Date

swift
RelevantContext.date(eventDate)

Date with Kind

DateKind provides an additional hint about the nature of the time relevance:

KindUse
.defaultGeneral time relevance
.scheduledA scheduled event (meeting, flight)
.informationalInformation relevant around a time (weather forecast)
swift
RelevantContext.date(meetingStart, kind: .scheduled)

Date Range

swift
// Using from/to
RelevantContext.date(from: startDate, to: endDate)

// Using DateInterval
RelevantContext.date(interval: dateInterval, kind: .scheduled)

// Using ClosedRange
RelevantContext.date(range: startDate...endDate, kind: .default)

Location-Based Relevance

Inferred Locations

The system infers certain locations from a person's routine. No coordinates needed.

swift
RelevantContext.location(inferred: .home)
RelevantContext.location(inferred: .work)
RelevantContext.location(inferred: .school)
RelevantContext.location(inferred: .commute)

Apply the location row in the Permissions table and check CLLocationManager.isAuthorizedForWidgetUpdates before returning clues.

Specific Region

swift
import CoreLocation

let region = CLCircularRegion(
    center: CLLocationCoordinate2D(latitude: 37.3349, longitude: -122.0090),
    radius: 500,
    identifier: "apple-park"
)
RelevantContext.location(region)

Point-of-Interest Category (26.0+ SDKs)

Indicate relevance near any location of a given category. Returns nil if the category is unsupported. The factory is SDK-available on Apple platforms 26.0+, but RelevanceKit clues still only affect Smart Stack behavior on watchOS.

swift
import MapKit

if let context = RelevantContext.location(category: .beach) {
    // Widget is relevant whenever the person is near a beach
}

Fitness and Sleep Relevance

Fitness

swift
// Relevant when activity rings are incomplete
RelevantContext.fitness(.activityRingsIncomplete)

// Relevant during an active workout
RelevantContext.fitness(.workoutActive)

Apply the exact fitness mapping in Permissions.

Sleep

swift
// Relevant around bedtime
RelevantContext.sleep(.bedtime)

// Relevant around wakeup
RelevantContext.sleep(.wakeup)

Apply the sleep mapping in Permissions.

Hardware Relevance

swift
// Relevant when headphones are connected
RelevantContext.hardware(headphones: .connected)

No special permission required.

Combining Signals

Return multiple WidgetRelevanceAttribute values in the WidgetRelevance array to make a widget relevant under several different conditions.

swift
func relevance() async -> WidgetRelevance<MyIntent> {
    var attributes: [WidgetRelevanceAttribute<MyIntent>] = []

    // Relevant during morning commute
    attributes.append(
        WidgetRelevanceAttribute(
            configuration: MyIntent(mode: .commute),
            context: .location(inferred: .commute)
        )
    )

    // Relevant at work
    attributes.append(
        WidgetRelevanceAttribute(
            configuration: MyIntent(mode: .work),
            context: .location(inferred: .work)
        )
    )

    // Relevant around a scheduled event
    for event in upcomingEvents {
        attributes.append(
            WidgetRelevanceAttribute(
                configuration: MyIntent(eventID: event.id),
                context: .date(event.date, kind: .scheduled)
            )
        )
    }

    return WidgetRelevance(attributes)
}

Order matters. Return relevance attributes ordered by priority. The system may use only a subset of the provided relevances.

Widget Integration

Relevant Widget with RelevanceConfiguration

swift
@available(watchOS 26, *)
struct MyRelevantWidget: Widget {
    var body: some WidgetConfiguration {
        RelevanceConfiguration(
            kind: "com.example.relevant-events",
            provider: MyRelevanceProvider()
        ) { entry in
            EventWidgetView(entry: entry)
        }
        .configurationDisplayName("Events")
        .description("Shows upcoming events when relevant")
    }
}

Associating with a Timeline Widget

When both a timeline widget and a relevant widget show the same data, use associatedKind to prevent duplicate cards. The system replaces the timeline widget card with relevant widget cards when they are suggested.

swift
RelevanceConfiguration(
    kind: "com.example.relevant-events",
    provider: MyRelevanceProvider()
) { entry in
    EventWidgetView(entry: entry)
}
.associatedKind("com.example.timeline-events")

Grouping

WidgetRelevanceGroup controls how the system groups widgets in the Smart Stack.

swift
// Opt out of default per-app grouping so each card appears independently
WidgetRelevanceAttribute(
    configuration: intent,
    group: .ungrouped
)

// Named group -- only one widget from the group appears at a time
WidgetRelevanceAttribute(
    configuration: intent,
    group: .named("weather-alerts")
)

// Default system grouping
WidgetRelevanceAttribute(
    configuration: intent,
    group: .automatic
)

RelevantIntent (Timeline Provider Path)

When using a timeline provider, also update RelevantIntentManager so the system has relevance data between timeline refreshes.

swift
import AppIntents

func updateRelevantIntents() async {
    let intents = events.map { event in
        RelevantIntent(
            MyWidgetIntent(event: event),
            widgetKind: "com.example.events",
            relevance: RelevantContext.date(from: event.start, to: event.end)
        )
    }
    try? await RelevantIntentManager.shared.updateRelevantIntents(intents)
}

Call this whenever relevance data changes -- not only during timeline refreshes.

Previewing Relevant Widgets

Use the entry, relevance-configuration, and full-provider recipes in Preview Recipes. Enable WidgetKit Developer Mode on the watch, test permissions granted and denied, and finish on a physical Apple Watch; see Testing Tips.

Common Mistakes

  • Using RelevanceKit API expecting iOS behavior. The API compiles on all platforms but only has effect on watchOS.
  • Duplicate Smart Stack cards. When offering both a timeline widget and a relevant widget for the same data, use .associatedKind(_:) to prevent duplication.
  • Not calling updateRelevantIntents. When using timeline providers, calling this only inside timeline() means the system has stale relevance data between refreshes. Update whenever data changes.
  • Ignoring nil from location(category:). This factory returns an optional. Not all MKPointOfInterestCategory values are supported.

Review Checklist

  • Routing: RelevanceKit is watchOS-effect-only, and WidgetKit, HealthKit, MapKit, and CoreLocation implementation remains in sibling scope.
  • Signals: contexts match the data model, attributes are priority-ordered, and every clue uses the exact Permissions-table setup.
  • Providers: entry, placeholder, relevance, and preview paths are present; location category optionals and widget-update authorization are handled.
  • Coordination: .associatedKind(_:) prevents duplicate cards, and updateRelevantIntents runs whenever timeline-provider data changes.
  • Testing: Developer Mode is enabled and previews cover display sizes plus granted and denied permission states.

References

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Relevancekit AI skill do?

Increase widget visibility on Apple Watch using RelevanceKit. Use when providing contextual relevance signals for watchOS widgets, declaring time-based or location-based relevance, combining multiple relevance providers, helping the system surface the right widget at the right time on watchOS 26, or routing mixed RelevanceKit/WidgetKit/HealthKit/MapKit Smart Stack scope.

Why use Relevancekit on TypingMind?

Because you install it once and use it with any model. Relevancekit is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Relevancekit in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/dpearson2699/swift-ios-skills/tree/main/skills/relevancekit. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Relevancekit?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Relevancekit?

As many as you like. As long as a model supports skills, you can use Relevancekit with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Relevancekit AI skill free?

It is published on GitHub by dpearson2699. Check the repository for licensing terms. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇