kysely

When to use

Use this skill when you need to:

• Build type-safe SQL queries with Kysely
• Define database interfaces and column types
• Write and run Kysely migrations
• Handle transactions, error recovery, and connection pooling
• Load related data efficiently (JSON aggregation, avoiding N+1)
• Use raw SQL safely within Kysely
• Test code that uses Kysely

How to use

Read individual rule files for detailed explanations and code examples:

• rules/setup.md - Installation, dialect config, database interface
• rules/type-safety.md - Database types, codegen, Generated, ColumnType
• rules/select-queries.md - SELECT patterns, joins, subqueries, CTEs
• rules/insert-update-delete.md - DML operations, returning, upserts
• rules/transactions.md - Transaction patterns, isolation levels, error recovery
• rules/migrations.md - Migration setup, up/down, schema builder API
• rules/raw-sql.md - sql template tag, parameterization, sql.raw dangers
• rules/relations.md - JSON aggregation, jsonArrayFrom, N+1 prevention
• rules/error-handling.md - Error types, constraint violations, executeTakeFirstOrThrow
• rules/performance.md - Connection pooling, pagination, batch operations
• rules/testing.md - Integration tests, transaction isolation, mocking
• rules/plugins.md - CamelCasePlugin, custom plugins, ParseJSONResultsPlugin

Core principles

• Type safety end-to-end: Define database interfaces, let Kysely infer query result types
• Narrow selects: Select only the columns you need — avoid selectAll()
• Parameterized queries: Always use the sql template tag — never concatenate user input
• Single instance: Create one Kysely instance per database, reuse it everywhere
• Thin abstraction: Kysely compiles to SQL 1:1 — learn the SQL it generates
• Expression builder first: Prefer eb callbacks over raw SQL for type-safe dynamic expressions