Create Persistence Adapter

Skill: create-persistence-adapter

Generates or updates up to four files per aggregate in adapter/outbound/db/:

• *Entity.kt — JPA entity mapping the aggregate to a database table
• *JpaRepository.kt — Spring Data JpaRepository interface
• *EntityMapper.kt — singleton object for entity ↔ domain conversion
• *PersistenceAdapter.kt — @Component implementing the outbound repository port

If any files already exist, the skill reads them, compares against the port interface and domain entity, and applies only what is missing or outdated.

Full code patterns: see references/full-pattern.md

Pattern

Trigger: "Create a persistence adapter for newsletter subscription"
Input:   services/example-service/src/main/kotlin/.../port/outbound/NewsletterSubscriptionRepository.kt
Output — four files under adapter/outbound/db/:
  NewsletterSubscriptionEntity.kt          – @Entity data class
  NewsletterSubscriptionJpaRepository.kt   – JpaRepository interface
  NewsletterSubscriptionEntityMapper.kt    – object with toDomain / toEntity
  NewsletterSubscriptionPersistenceAdapter.kt – @Component implementing the port

What is a Persistence Adapter?

A persistence adapter is an outbound adapter. It acts as the interaction layer between your application and the database. Your code calls it via a port interface. It handles the translation between domain objects and the underlying storage mechanism (JPA/Spring Data).

IMPORTANT

• Implement every method defined in the outbound repository port interface
• find throws NoSuchElementException when the entity is not found; search returns null
• All entity ↔ domain mapping goes through the singleton mapper object — never inline conversion.
• The mapper is a Kotlin object (singleton), not a class or a companion object
• Column names use snake_case; the @Entity(name = ...) value matches the database table name
• Enum fields use @Enumerated(EnumType.STRING) — never store the ordinal
• The JPA repository extends JpaRepository<EntityClass, UUID>
• It adds adds one findBy<IdField> method returning a nullable entity for kotlin-friendly find

Instructions

Step 1 – Resolve the input source

Two possible inputs:

Option A – Repository port file path provided (a .kt file under port/outbound/):

1. Read the file. Extract: interface name, all method signatures (names, parameter types, return types).
2. Derive the aggregate name by stripping the Repository suffix (e.g. NewsletterSubscriptionRepository → NewsletterSubscription).
3. Find the domain aggregate: Glob **/domain/<AggregateName>.kt. Read it to discover all fields.
4. If you can't find the matching aggregate or port pause the execution
5. Then ask the developer to provide correct information

Option B – No argument provided:

• Search: Glob **/application/port/outbound/**/*.kt.
• Filter the files to include only ones that could match as a port for a repository.
• If no files are found, inform the developer that no outbound port files were discovered, ask them to provide the port file path directly, then proceed as Option A. -
• If files were found list those files and ask the user which one to scaffold.
• Proceed as Option A once the user selects a file.

Step 2 – Determine the target package and source root

Derive the base package from the repository port package (e.g. io.miragon.example.application.port.outbound → base: io.miragon.example).

Adapter package: <base>.adapter.outbound.db.

Determine the source root from the port file path (e.g. services/example-service/src/main/kotlin/).

Target directory: <source-root>/<base-package-path>/adapter/outbound/db/

Step 3 – Check existing files and determine should-state for each

Derive the four class names from the aggregate name (e.g. NewsletterSubscription):

• Entity: NewsletterSubscriptionEntity
• JPA Repository: NewsletterSubscriptionJpaRepository
• Mapper: NewsletterSubscriptionEntityMapper
• Adapter: NewsletterSubscriptionPersistenceAdapter

Check which of these files already exist at the target directory. For each file, determine its should-state as follows:

3.1 Entity (<AggregateName>Entity.kt)

Read the domain aggregate class. For each field:

• Map domain value objects to their primitive type (e.g. Email(val value: String) → String, SubscriptionId(val value: UUID) → UUID)
• Derive the column name as snake_case of the field name (e.g. registrationDate → registration_date, newsletterId → newsletter_id)
• Mark enum fields with @Enumerated(EnumType.STRING)
• Mark the ID field with @Id and @Column(name = "<id-column>", nullable = false)

Derive the table name as snake_case of the aggregate class name (e.g. NewsletterSubscription → newsletter_subscription).

Should-state: data class with @Entity(name = "table_name"), one column per aggregate field, correct annotations.

If file exists: compare fields against the domain aggregate. Add missing fields, correct outdated column names or annotations. Preserve any custom annotations already present.

If file does not exist: generate from the should-state above.

3.2 JPA Repository (<AggregateName>JpaRepository.kt)

Should-state: interface extending JpaRepository<EntityClass, UUID> with exactly one findBy<IdField>(id: UUID): EntityClass? query method.

If the file exists: check if any query methods are missing based on the port interface. Add them.

If file does not exist: generate from the should-state above.

3.3 Mapper (<AggregateName>EntityMapper.kt)

Should-state: object with two functions:

• toDomain(entity: EntityClass): DomainClass — wraps each primitive in its domain value object
• toEntity(domain: DomainClass): EntityClass — extracts .value from each domain value object

If a file exists: check for missing field mappings and add them.

If file does not exist: generate from the should-state above.

3.4 Adapter (<AggregateName>PersistenceAdapter.kt)

Should-state: @Component class implementing the repository port, constructor-injected JPA repository, all port methods delegating through the mapper:

• find → throws NoSuchElementException when not found
• search → returns null when not found
• save → delegates to repository.save(...)
• delete → delegates to repository.deleteById(id.value)

If file exists: compare against all port methods. Add any missing ones, preserving existing business logic.

If file does not exist: generate from the should-state above.

Step 4 – Write all files

Apply all creations and updates identified in Step 3. Use references/full-pattern.md as a concrete reference.

Step 5 – Report

List each file created, updated, or skipped (if nothing changed) and remind the developer to:

1. Add a database migration (Flyway/Liquibase) if a new table is introduced
2. Run /test-persistence-adapter to generate unit tests for the adapter
3. Wire the adapter into the application service via the port interface (automatic via Spring if the package is within the base component-scan path)