migrate

Database Migrate

Generate migration files for database schema changes.

Load the database-designer skill for migration strategy and schema versioning patterns.

Usage

/migrate <description> [--type sql|knex|typeorm|drizzle]

Arguments

Parse $ARGUMENTS to extract:

• description (required) — What the migration does (e.g., "add-user-roles", "create-orders-table")
• --type (optional) — Migration framework. Auto-detected if not specified.

Instructions

1. Detect Migration Framework

ls prisma/schema.prisma 2>/dev/null           # Prisma → use prisma migrate
ls knexfile.* migrations/ 2>/dev/null          # Knex
ls ormconfig.* src/migrations/ 2>/dev/null     # TypeORM
ls drizzle.config.* 2>/dev/null               # Drizzle
ls alembic.ini alembic/ 2>/dev/null           # Alembic (Python)

If no framework detected, generate raw SQL migration files.

2. Understand Current Schema

• Read the current schema (ORM files, existing migrations, or MCP introspection)
• Understand what exists before generating changes

3. Generate Migration

Raw SQL (default when no ORM): Create timestamped file in migrations/:

migrations/
  YYYYMMDDHHMMSS_<description>.up.sql
  YYYYMMDDHHMMSS_<description>.down.sql

The UP migration should:

• Create tables, add columns, create indexes, etc.
• Use IF NOT EXISTS where appropriate
• Include comments explaining the change

The DOWN migration should:

• Reverse every change in the UP migration
• Use IF EXISTS for safety

Knex: Generate a migration file using the Knex pattern:

npx knex migrate:make <description>

Then populate the generated file with the schema changes.

TypeORM:

npx typeorm migration:create src/migrations/<Description>

Drizzle:

npx drizzle-kit generate

4. Review Before Applying

Present the generated migration to the user:

• Show the full migration code
• Explain what it does
• Warn about any destructive operations (DROP, ALTER column type)
• Note if it requires a data migration (not just schema)
• Never run the migration automatically — let the user decide