kysely-codegen

Use the Kysely CamelCasePlugin for generated table column names. When enabled, column names like created_at become createdAt in the generated types.

Specifies which parser to use for PostgreSQL date/time values.

• timestamp — Dates are typed as Date (default)
• string — Dates are typed as string

Sets the default schema(s) for the database connection. Tables in these schemas are generated without a schema prefix.

Defaults to ["public"] for PostgreSQL and [] for other dialects.

Sets the SQL dialect. If not provided, it is inferred from the connection string URL.

• postgres — PostgreSQL
• mysql — MySQL / MariaDB
• mssql — Microsoft SQL Server
• sqlite — SQLite (via better-sqlite3)
• bun-sqlite — SQLite via Bun's built-in SQLite
• kysely-bun-sqlite — SQLite via kysely-bun-sqlite
• worker-bun-sqlite — SQLite via Bun worker
• libsql — libSQL (Turso)

Controls whether to generate types for PostgreSQL domains. Only relevant when using the postgres dialect.

Path to an environment file (e.g. .env) to load before reading the connection string. Useful when DATABASE_URL is defined in an env file.

A glob pattern to exclude tables from code generation. Tables matching this pattern are skipped.

A glob pattern to include only matching tables in code generation. Tables not matching this pattern are skipped.

Sets the terminal log level.

• debug — All messages
• info — Informational and above
• warn — Warnings and errors (default)
• error — Errors only
• silent — No output

Specifies which parser to use for PostgreSQL numeric / decimal values.

• string — Typed as string (default, preserves precision)
• number — Typed as number (may lose precision for large values)
• number-or-string — Typed as number | string

Sets the output file path for the generated type definitions. When set to null, uses the default path inside node_modules.

Specifies type overrides for specific table columns. Allows you to manually set the TypeScript type for columns where the inferred type is not ideal.

Includes PostgreSQL table partitions in the generated code. Only relevant when using the postgres dialect.

Prints the generated output to the terminal (stdout) instead of writing to a file. Useful for previewing output or piping to other tools.

Generates TypeScript enum declarations for PostgreSQL enum types instead of union types. Only relevant when using the postgres dialect.

• false — Generate union types like "a" | "b" (default)
• true — Generate runtime enums with default naming
• "pascal-case" — Generate enums with PascalCase member names
• "screaming-snake-case" — Generate enums with SCREAMING_SNAKE_CASE member names

Singularizes generated type aliases (e.g. Users becomes User).

• false — No singularization (default)
• true — Use built-in singularization rules
• {...} — A record of regex pattern to replacement string for custom singularization rules

Skips adding the // This file was generated by kysely-codegen comment at the top of the generated file.

Uses TypeScript's import type syntax in the generated code (e.g. import type { ColumnType } from 'kysely' instead of import { ColumnType } from 'kysely').

The database connection string URL. If not provided, reads from the DATABASE_URL environment variable.

Examples for each dialect:

• PostgreSQL: postgres://user:pass@localhost:5432/mydb
• MySQL: mysql://user:pass@localhost:3306/mydb
• SQLite: C:/path/to/db.sqlite or :memory:
• libSQL: libsql://your-db.turso.io
• MSSQL: Server=localhost;Database=mydb;User Id=sa;Password=pass;

Verifies that the generated types are up-to-date without writing any files. Exits with a non-zero code if the output file would change. Useful in CI pipelines.