Stabilizing Compose Types logo

Stabilizing Compose Types

Community
skydoves
stabilizing-compose-types

Use this skill to fix unstable Jetpack Compose types once a stability diagnosis has identified them. Covers the three-tier strategy — make the type truly stable with val plus immutable fields, mark with @Immutable or @Stable when the source is owned, and use stabilityConfigurationFiles for third-party or Java types. Explains the compiler-level difference between @Immutable and @Stable (static expression promotion), kotlinx.collections.immutable for List/Set/Map parameters, and the StableHolder wrapper escape hatch. Use when the developer asks how to stabilize a User class, a List parameter, java.time.LocalDateTime, a Flow parameter, or when the compiler report shows unstable params and the developer wants the fix. The diagnostic step lives in a sibling skill.

Overview

Publisherskydoves
Repositorycompose-performance-skills
Skill namestabilizing-compose-types
Stars
515
Forks
19
Bundled files
1
LicenseApache-2.0
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.

  • 1 bundled files

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

  • Open source

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

Installation

Install the Stabilizing Compose Types 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/skydoves/compose-performance-skills/tree/main/stability/stabilizing-compose-types
  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/skydoves/compose-performance-skills.git /tmp/compose-performance-skills
mkdir -p .claude/skills
cp -r /tmp/compose-performance-skills/stability/stabilizing-compose-types .claude/skills/stabilizing-compose-types
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Stabilizing Compose Types 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 Stabilizing Compose Types 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 Stabilizing Compose Types 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.

Stabilizing Compose Types — Three Tiers, In Order

Once a diagnosis (see ../diagnosing-compose-stability/SKILL.md) has named the unstable types, this skill walks Claude through fixing them. The strategy is a strict three-tier waterfall: (1) make the type truly stable by structural rewrite (val + immutable fields) — no annotation needed; (2) annotate with @Immutable or @Stable when the source is owned and the contract is honored; (3) use stabilityConfigurationFile for third-party or Java types. DO NOT invert this order — annotations are a contract, not a magic spell.

When to use this skill

  • The compiler report (composables.txt / classes.txt) shows one or more unstable parameters.
  • The developer asks how to stabilize a domain class, a List<Foo> parameter, java.time.LocalDateTime, or a third-party type.
  • The developer mentions @Immutable, @Stable, compose-stable-marker, compose-runtime-annotation, kotlinx.collections.immutable, stabilityConfigurationFiles, or stability_config.conf.
  • A @TraceRecomposition log shows recomposition happening because an argument is allocated fresh every recomposition.

When NOT to use this skill

  • The unstable types are not yet identified — run ../diagnosing-compose-stability/SKILL.md first.
  • The symptom is a wrong-phase state read (Modifier.alpha(state.value)); use ../../recomposition/deferring-state-reads/SKILL.md.
  • The symptom is a derivedStateOf problem; use ../../recomposition/choosing-derivedstateof/SKILL.md.
  • A Flow<T> parameter is the cause; the fix is to collect upstream — see ../../side-effects/collecting-flows-safely/SKILL.md rather than annotating Flow as stable.

Prerequisites

  • ../diagnosing-compose-stability/SKILL.md has been run; the developer has a concrete list of unstable types.
  • Kotlin 2.0.0+ with org.jetbrains.kotlin.plugin.compose applied. Strong Skipping is on by default.
  • For pure-Kotlin / data modules without compose-runtime: either androidx.compose.runtime:runtime-annotation (official) or com.github.skydoves:compose-stable-marker (legacy) on the classpath. Both expose @Stable / @Immutable / @StableMarker without dragging in the full runtime.
  • For tier-3 fixes: Compose Compiler 1.5.5+ for stabilityConfigurationFiles DSL support (plural; the older singular stabilityConfigurationFile property is deprecated, see footnote below).

Workflow — decision tree

  • 1. Do we own the source of the unstable type? If yes, use tier 1 or tier 2. If no (Java stdlib, third-party SDK), jump to tier 3.

  • 2. Tier 1 — restructure to truly stable. Can every property be a val of an already-stable type? If yes, MUST make that change first. No annotation is needed and no contract is implied. The compiler will infer stability automatically.

kotlin
// WRONG
data class Snack(
    var name: String,
    val tags: Set<String>,
)
// WRONG because: `var name` is an observable that does not notify Compose, and `Set<String>` is the standard library interface so its stability is Unknown — the data class is unstable on two axes.
kotlin
// RIGHT
@Immutable
data class Snack(
    val name: String,
    val tags: ImmutableSet<String>,
)
  • 3. Is the unstable property a collection? Replace kotlin.collections.List, Set, Map with kotlinx.collections.immutable.ImmutableList, ImmutableSet, ImmutableMap. Build with persistentListOf(), persistentSetOf(), persistentMapOf() factories or .toImmutableList() adapters. The kotlinx-collections-immutable artifact ships a known-stable bitmask 0b1 recognized by the Compose Compiler. PREFERRED: prefer this over whitelisting kotlin.collections.* in stability_config.conf because the type system enforces immutability.

  • 4. Is the source a pure-Kotlin / data module that does not depend on compose-runtime? Use androidx.compose.runtime:runtime-annotation (official, recommended) or com.github.skydoves:compose-stable-marker (legacy). Both let the data module annotate types without pulling in the full Compose runtime. Skydoves hot take #6: pure-Kotlin / data modules can use compose-stable-marker (or newer official compose-runtime-annotation) without pulling in full compose-runtime.

kotlin
// shared data module — build.gradle.kts
dependencies {
    // Official, preferred:
    compileOnly("androidx.compose.runtime:runtime-annotation:<version>")
    // Or legacy skydoves:
    // implementation("com.github.skydoves:compose-stable-marker:<version>")
}
  • 5. Tier 2 — choose @Immutable or @Stable. If every property is val AND every nested value is itself immutable AND equals() is structural, use @Immutable. Otherwise — for types whose values can change but whose mutations are observed by Compose (e.g. holders containing MutableState) — use @Stable.

The compiler treats @Immutable more aggressively than @Stable. For @Immutable types, when every constructor argument at a call site is a compile-time constant (e.g. literal numbers, top-level vals of stable types, or other @Immutable-with-constant-args), the compiler performs static expression promotion: the constructed instance is hoisted into a singleton, the lambda capture sites are de-duplicated, and the @Immutable parameter is marked @static in composables.txt. Most articles describe @Stable and @Immutable as interchangeable — they are not.

kotlin
// RIGHT — @Immutable: every property val, every property type immutable
@Immutable
data class ThemeColors(
    val primary: Color,
    val onPrimary: Color,
    val background: Color,
)

// RIGHT — @Stable: mutable but mutations notify Compose via Snapshot
@Stable
class CartState {
    var total: Money by mutableStateOf(Money.ZERO)
    val lines: SnapshotStateList<Line> = mutableStateListOf()
}
  • 6. Tier 3 — third-party or Java types. Use stabilityConfigurationFiles (plural). Create stability_config.conf at the project root listing exact-class or wildcard patterns; wire it into the composeCompiler { } block via stabilityConfigurationFiles.add(...). Full grammar lives in references/stability-config-syntax.md.
kotlin
// build.gradle.kts (root or composable module)
composeCompiler {
    stabilityConfigurationFiles.add(
        rootProject.layout.projectDirectory.file("stability_config.conf")
    )
}

Footnote — legacy form. Older projects may still wire the file via the singular property stabilityConfigurationFile = .... That property is @Deprecated("Use the stabilityConfigurationFiles option instead") in modern Compose Compiler Gradle plugin releases — prefer the plural stabilityConfigurationFiles.add(...) shown above.

text
# stability_config.conf — opt-in contract
java.time.LocalDateTime
java.time.LocalDate
kotlin.collections.List
kotlin.collections.Set
kotlin.collections.Map
com.example.thirdparty.**
  • 7. Nothing in tiers 1-3 fits? Wrap in a @Stable class StableHolder<T>(val item: T). This is the escape hatch — the holder's identity is stable, equality delegates to item, and Compose can skip on it.
kotlin
@Stable
class StableHolder<T>(val item: T) {
    override fun equals(other: Any?) = other is StableHolder<*> && other.item == item
    override fun hashCode() = item?.hashCode() ?: 0
}
  • 8. Flow<T> parameters — DO NOT annotate as stable. Skydoves hot take #4: Flow parameters are unstable. Don't pass flows to composables; collect them in a ViewModel or with collectAsStateWithLifecycle.
kotlin
// WRONG
@Composable
fun Feed(items: Flow<List<Item>>) { /* ... */ }
// WRONG because: Flow has no observable identity; the composable cannot skip on it, and a downstream `collectAsState` here detaches from lifecycle.
kotlin
// RIGHT — collect upstream and pass the resolved value
@Composable
fun FeedRoute(viewModel: FeedViewModel = viewModel()) {
    val items by viewModel.items.collectAsStateWithLifecycle()
    Feed(items = items)
}

@Composable
fun Feed(items: ImmutableList<Item>) { /* ... */ }
  • 9. Re-run the diagnostic skill to verify the previously unstable params are now stable or runtime.

Patterns

Pattern: data class with var and Set field

kotlin
// WRONG
data class Snack(
    var name: String,
    val tags: Set<String>,
)
// WRONG because: `var` blocks compile-time stability inference; `Set<String>` is an interface with unknown implementations.
kotlin
// RIGHT
import kotlinx.collections.immutable.ImmutableSet
import kotlinx.collections.immutable.persistentSetOf

@Immutable
data class Snack(
    val name: String,
    val tags: ImmutableSet<String> = persistentSetOf(),
)

Pattern: List<Item> parameter on a composable

kotlin
// WRONG
@Composable
fun ItemList(items: List<Item>) { /* ... */ }
// WRONG because: `kotlin.collections.List` is an interface; the compiler cannot prove every implementation is immutable, so the parameter is `unstable` and skipping is disabled (or pinned to `===` under Strong Skipping, which still fails on every fresh allocation).
kotlin
// RIGHT
import kotlinx.collections.immutable.ImmutableList

@Composable
fun ItemList(items: ImmutableList<Item>) { /* ... */ }

Producer side:

kotlin
val items: ImmutableList<Item> = repository.items().toImmutableList()

Pattern: java.time.LocalDateTime flagged unstable

The Java time types are separately compiled, no annotation, so the inference falls through to "unknown". Tier 3 is the right answer.

text
# stability_config.conf
java.time.LocalDateTime
java.time.LocalDate
java.time.Instant
java.time.ZonedDateTime
java.time.Duration
kotlin
// WRONG — annotate a wrapper to "force" stability without honoring the contract
@Stable
class DateWrapper(var date: LocalDateTime)
// WRONG because: `var` field, no Snapshot notification — the @Stable contract is broken; recompositions will be silently missed.
kotlin
// RIGHT — whitelist the immutable JDK type via stability config
// stability_config.conf:  java.time.LocalDateTime
@Immutable
data class Order(val placedAt: LocalDateTime, val total: Money)

Pattern: Flow<T> parameter on a composable

kotlin
// WRONG
@Composable
fun Detail(productFlow: Flow<Product>) {
    val product by productFlow.collectAsState(initial = null)
    /* ... */
}
// WRONG because: `Flow` is a cold producer with no observable identity; the composable can never skip on it, and `collectAsState` (no `WithLifecycle`) keeps collecting in the background.
kotlin
// RIGHT — hoist collection to the route
@Composable
fun DetailRoute(viewModel: DetailViewModel = viewModel()) {
    val product by viewModel.product.collectAsStateWithLifecycle()
    Detail(product = product)
}

@Composable
fun Detail(product: Product?) { /* ... */ }

Cross-reference: ../../side-effects/collecting-flows-safely/SKILL.md.

Pattern: inline composable wrap as a "fix"

kotlin
// WRONG — wrapping a Row in an extracted composable to "force" skippability
@Composable
fun ItemRow(item: Item) {
    Row { /* ... */ }
}
// WRONG because: `Row`/`Column`/`Box` are inline composables — they are NOT restartable/skippable to begin with (skydoves hot take #3). Wrapping them creates a new restart scope, which can change behavior unpredictably without addressing the root unstable parameter.
kotlin
// RIGHT — make the parameter stable, leave the inline composable alone
@Immutable
data class Item(val id: Long, val name: String)

@Composable
fun ItemRow(item: Item) {
    Row { /* ... */ }
}

Pattern: @Immutable versus @Stable — pick the stronger one

kotlin
// SUFFICIENT but suboptimal
@Stable
data class ThemeColors(
    val primary: Color,
    val onPrimary: Color,
)
// Sufficient because: the compiler still treats it as stable.
// Suboptimal because: @Immutable would additionally enable static expression promotion when ThemeColors is constructed with all-constant args.
kotlin
// PREFERRED
@Immutable
data class ThemeColors(
    val primary: Color,
    val onPrimary: Color,
)

In composables.txt, calls like ThemeColors(Color(0xFFFFFFFF), Color(0xFF000000)) will then be reported as stable @static themeColors: ThemeColors — the constructed instance is hoisted to a singleton. MUST prefer @Immutable whenever the type qualifies.

Mandatory rules

  • MUST NOT annotate a type as @Stable or @Immutable unless the contract is honored. Skydoves hot take #2: stability config is a contract, not a magic spell — break it and recompositions are silently missed.
  • MUST prefer @Immutable over @Stable whenever every property is a val of an already-immutable type. The compiler emits stronger optimizations (static expression promotion, lambda-singleton, compile-time default eval) for @Immutable.
  • MUST prefer kotlinx.collections.immutable over annotating a mutable collection as stable. The compiler ships a known-stable bitmask for these types.
  • MUST NOT wrap inline composables (Row/Column/Box) to "force" skippability. Inline composables are not restartable in the first place; wrapping changes scoping without addressing the root cause.
  • MUST NOT pass Flow<T> as a composable parameter. Collect upstream with collectAsStateWithLifecycle.
  • MUST re-run ../diagnosing-compose-stability/SKILL.md after applying fixes; the previously unstable params MUST now report stable or runtime.
  • PREFERRED: stabilityConfigurationFiles (plural; stabilityConfigurationFiles.add(...)) over scattering annotations across modules the team does not own.
  • PREFERRED: in pure-Kotlin / data modules, use androidx.compose.runtime:runtime-annotation (official) or com.github.skydoves:compose-stable-marker (legacy) so the annotation is available without a full compose-runtime dependency.

Verification

  • Re-run release-mode build with composeCompiler { reportsDestination = ... } enabled.
  • In composables.txt, every previously unstable parameter is now stable or runtime.
  • In classes.txt, every fixed class is now stable class … or runtime stable class … — no unstable remains for the targeted types.
  • If stabilityConfigurationFiles was used, every entry has been verified to honor the contract (no hidden mutation, structural equals, no observable that bypasses Snapshot).
  • @TraceRecomposition (skydoves/compose-stability-analyzer) on the previously hot composable shows recomposition counts dropping to the expected number per state change.
  • No @Stable / @Immutable annotation has been added to a type that still contains a var or an unstable nested type.

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 Stabilizing Compose Types AI skill do?

Use this skill to fix unstable Jetpack Compose types once a stability diagnosis has identified them. Covers the three-tier strategy — make the type truly stable with val plus immutable fields, mark with @Immutable or @Stable when the source is owned, and use stabilityConfigurationFiles for third-party or Java types. Explains the compiler-level difference between @Immutable and @Stable (static expression promotion), kotlinx.collections.immutable for List/Set/Map parameters, and the StableHolder wrapper escape hatch. Use when the developer asks how to stabilize a User class, a List parameter,...

Why use Stabilizing Compose Types on TypingMind?

Because you install it once and use it with any model. Stabilizing Compose Types 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 Stabilizing Compose Types in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/skydoves/compose-performance-skills/tree/main/stability/stabilizing-compose-types. 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 Stabilizing Compose Types?

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 Stabilizing Compose Types?

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

Is the Stabilizing Compose Types AI skill free?

Yes. It is published on GitHub by skydoves under the Apache-2.0 license. 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 👇