agentleFS
rsicarellirsicarelli/fakt/.claude/skills/feature-option/SKILL.md

feature-option

Skill67 starsChanged 47 days agoApache-2.0View on GitHub

Guides step-by-step through all 11 touchpoints for adding a new @Fake annotation option (like mutability or callHistory). Covers annotation enum, CLI processor, Gradle plugin, FIR metadata, FIR translator, the shared FakeDeclaration model, and code generators. Use when adding a new option to @Fake, adding a new enum to the annotation, making something configurable per-interface, implementing a new compiler plugin feature flag, adding a new Gradle extension property, or extending the three-tier resolution (annotation → plugin → default). Make sure to use this skill whenever a change touches more than one layer of the option pipeline, even for small additions like a new enum value.

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 2 are distinctive, and 6 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 6

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use AskUserQuestion to collect:

    line 15
  2. Very specific:

    Add parameter with DEFAULT default:

    line 47
  3. Very specific:

    Add Property<Boolean> with convention default:

    line 65
  4. Very specific:

    Add field + read in load():

    line 111
  5. Distinctive:

    Add private extraction method in both:

    line 142
  6. Very specific:

    Call in analyzeMetadata() and pass to ValidatedFakeInterface/ValidatedFakeClass constructor.

    line 167
  7. Very specific:

    Add the resolved boolean to the sealed interface and override it in both Interface and Class:

    line 193
  8. Distinctive:

    Every option follows this resolution:

    line 231

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Feature Option Guide

Adds a new option to the @Fake annotation, following the confirmed pattern from mutability and callHistory.

Instructions

1. Gather Option Details

Use AskUserQuestion to collect:

  • Option name (e.g., threadSafety)
  • Enum values — always DEFAULT plus two semantically opposite values (e.g., ENABLED/DISABLED)
  • Plugin-level default — the boolean value when annotation says DEFAULT (e.g., true or false)
  • What it controls — which generators are affected

2. Walk Through All 11 Layers

Guide the implementation in this exact order. Each layer depends on the previous.


Layers 1-2 define the user-facing API that developers interact with — the enum they import and the annotation parameter they set.

Layer 1 — Annotation Enum (new file)

File: annotations/src/commonMain/kotlin/com/rsicarelli/fakt/{OptionName}Mode.kt

public enum class {OptionName}Mode {
    DEFAULT,   // Follow plugin-level setting
    {VALUE_A}, // Explicit enable
    {VALUE_B}, // Explicit disable
}

Pattern: MutabilityMode(DEFAULT, IMMUTABLE, MUTABLE), CallHistoryMode(DEFAULT, ENABLED, DISABLED).

Layer 2 — @Fake Annotation Parameter

File: annotations/src/commonMain/kotlin/com/rsicarelli/fakt/Fake.kt

Add parameter with DEFAULT default:

public annotation class Fake(
    val callHistory: CallHistoryMode = CallHistoryMode.DEFAULT,
    val mutability: MutabilityMode = MutabilityMode.DEFAULT,
    val {optionName}: {OptionName}Mode = {OptionName}Mode.DEFAULT, // NEW
)

Naming note: Real field names are descriptive and may not follow the template literally. E.g., mutability uses enableMutableFakes (not enableMutability) and generateMutableBehaviors (not generateMutability). Choose names that best describe the feature's effect.

Layers 3-4 wire the option through the Gradle build system to the compiler — this is how users set project-wide defaults in build.gradle.kts.

Layer 3 — Gradle Plugin Extension

File: gradle-plugin/src/main/kotlin/com/rsicarelli/fakt/gradle/FaktPluginExtension.kt

Add Property<Boolean> with convention default:

public val enable{OptionName}: Property<Boolean> =
    objects.property(Boolean::class.java).convention({defaultValue})

Layer 4 — Gradle Subplugin (passes option to compiler)

File: gradle-plugin/src/main/kotlin/com/rsicarelli/fakt/gradle/FaktGradleSubplugin.kt

In applyToCompilation, add SubpluginOption:

add(SubpluginOption(key = "enable{OptionName}", value = extension.enable{OptionName}.get().toString()))

Layers 5-6 register the option with compiler plugin infrastructure — the CLI processor receives the value from Gradle, and FaktOptions makes it available throughout the compiler.

Layer 5 — Compiler CLI Processor

File: compiler/src/main/kotlin/com/rsicarelli/fakt/compiler/core/config/FaktCommandLineProcessor.kt

Three additions in companion object + one in processOption:

// Companion:
val ENABLE_{OPTION}_KEY = CompilerConfigurationKey<Boolean>("fakt.enable{OptionName}")
val ENABLE_{OPTION}_OPTION = CliOption(
    optionName = "enable{OptionName}",
    valueDescription = "true|false",
    description = "Enable {description} by default (default: {defaultValue})",
    required = false,
)

// Add to pluginOptions list:
ENABLE_{OPTION}_OPTION

// In processOption when-branch:
"enable{OptionName}" -> configuration.put(ENABLE_{OPTION}_KEY, value.toBoolean())

Layer 6 — FaktOptions Data Class

File: compiler/src/main/kotlin/com/rsicarelli/fakt/compiler/core/config/FaktOptions.kt

Add field + read in load():

data class FaktOptions(
    // ...existing...
    val enable{OptionName}Default: Boolean = {defaultValue},
)

// In load():
val enable{OptionName} = configuration.get(FaktCommandLineProcessor.ENABLE_{OPTION}_KEY) ?: {defaultValue}
return FaktOptions(/* ...existing... */, enable{OptionName}Default = enable{OptionName})

Layers 7-8 extract the annotation value at compile time during the FIR phase — mirror enums avoid a runtime dependency on the annotations module, and checkers read the actual @Fake parameter value from source code.

Layer 7 — FIR Mirror Enum + Validated Metadata

File: codegen-runtime/src/main/kotlin/com/rsicarelli/fakt/compiler/fir/metadata/FirFakeMetadata.kt

// Mirror enum (avoids runtime dep on annotations):
enum class Fir{OptionName}Mode { DEFAULT, {VALUE_A}, {VALUE_B} }

// Add to BOTH ValidatedFakeInterface and ValidatedFakeClass:
val {optionName}Mode: Fir{OptionName}Mode = Fir{OptionName}Mode.DEFAULT,

Layer 8 — FIR Checkers (annotation extraction)

Files: compiler/.../fir/checkers/FakeInterfaceChecker.kt + FakeClassChecker.kt

Add private extraction method in both:

private fun extract{OptionName}Mode(declaration: FirClass, session: FirSession): Fir{OptionName}Mode {
    val fakeAnnotation = declaration.annotations.find { annotation ->
        annotation.toAnnotationClassIdSafe(session) == FAKE_ANNOTATION_CLASS_ID
    } ?: return Fir{OptionName}Mode.DEFAULT

    val arg = fakeAnnotation.argumentMapping.mapping.entries
        .find { it.key.asString() == "{optionName}" }?.value

    return when (arg) {
        is FirPropertyAccessExpression -> {
            val enumEntry = arg.calleeReference.toResolvedEnumEntrySymbol()
            when (enumEntry?.name?.asString()) {
                "{VALUE_A}" -> Fir{OptionName}Mode.{VALUE_A}
                "{VALUE_B}" -> Fir{OptionName}Mode.{VALUE_B}
                else -> Fir{OptionName}Mode.DEFAULT
            }
        }
        else -> Fir{OptionName}Mode.DEFAULT
    }
}

Call in analyzeMetadata() and pass to ValidatedFakeInterface/ValidatedFakeClass constructor.

Layers 9-10 resolve the option and carry the result to code generation. New generation behaviour goes on the FIR path (see CLAUDE.md); the resolver implements the three-tier resolution (annotation → plugin → default) and the shared FakeDeclaration model carries the final boolean that generators consume.

Layer 9 — Resolver + FIR Translator

Files: codegen-runtime/.../compiler/fir/metadata/FirFakeMetadata.kt + compiler/.../fir/generation/FirToFakeDeclarationTranslator.kt + compiler/.../fir/generation/FirFakeEmitter.kt

a) Resolver next to the mirror enum (pattern: resolveCallHistory, resolveMutability):

fun Fir{OptionName}Mode.resolve{OptionName}(pluginDefault: Boolean): Boolean =
    when (this) {
        Fir{OptionName}Mode.{VALUE_A} -> true
        Fir{OptionName}Mode.{VALUE_B} -> false
        Fir{OptionName}Mode.DEFAULT -> pluginDefault
    }

b) In both toFakeInterface() and toFakeClass(), add an enable{OptionName}Default: Boolean parameter and set generate{OptionName} = {optionName}Mode.resolve{OptionName}(enable{OptionName}Default).

c) In both FirFakeEmitter.emit() overloads, pass sharedContext.options.enable{OptionName}Default.

Layer 10 — Shared Model (final boolean)

File: codegen-runtime/.../codegen/analysis/FakeDeclaration.kt

Add the resolved boolean to the sealed interface and override it in both Interface and Class:

val generate{OptionName}: Boolean

The legacy compiler/.../ir/analysis/IrToFakeDeclarationTranslator.kt builds the same model, so it needs the argument to compile until #150 removes it; add no other IR-side logic.

Layer 11 is where the option actually affects generated code — generators read the resolved boolean and conditionally emit different Kotlin source.

Layer 11 — Code Generators (consume the boolean)

Modify generators based on what the option controls:

  • ImplementationGenerator.kt — pass decl.generate{OptionName} to generateCompleteFake()
  • ConfigurationDslGenerator.kt — conditionally generate DSL elements
  • FakeGenerator.kt / extensions — conditional code generation

3. Validate Implementation

Checklist:

  • Enum file created in annotations/ with DEFAULT + two values
  • @Fake annotation has new parameter with DEFAULT default
  • Gradle extension has Property<Boolean> with convention
  • Subplugin passes option via SubpluginOption
  • CLI processor has KEY, OPTION, pluginOptions entry, processOption branch
  • FaktOptions has field + load() reads it
  • FIR mirror enum + fields in both ValidatedFakeInterface and ValidatedFakeClass
  • Both checkers have extract{OptionName}Mode() + call in analyzeMetadata()
  • resolve{OptionName}() next to the mirror enum in FirFakeMetadata.kt
  • toFakeInterface() and toFakeClass() set the resolved boolean; both FirFakeEmitter.emit() overloads pass the plugin default
  • FakeDeclaration (and its Interface/Class) has generate{OptionName}: Boolean
  • Generators consume the boolean
  • make publish-local && make test-sample passes
  • make format applied

4. Three-Tier Resolution Pattern

Every option follows this resolution:

Annotation value (per-interface)
    ↓ if DEFAULT
Plugin setting (build.gradle.kts)
    ↓ if not set
Hardcoded default (in FaktOptions)

This allows: global default via Gradle, per-interface override via annotation, sensible fallback.

Related Skills

  • sample-scaffolder — add a sample that demonstrates the option
  • codegen — understand code generation pipeline for Layer 11

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.