drizzle_skill

Overview

This skill is a practical Drizzle ORM schema and database guide for TypeScript projects. It codifies naming, column patterns, timestamp helpers, index usage, and migration best practices to keep schemas consistent and migrations safe. Use it when defining tables, creating migrations, or inferring types from Drizzle tables.

How this skill works

It inspects Drizzle schema files in src/database/schemas/* and provides guidance on table definitions, helper utilities, and common patterns like junction tables. It explains how to define primary keys, foreign keys, timestamps, indexes, and how to generate and harden migrations for PostgreSQL with strict dialect settings. It also shows how to infer TypeScript types from Drizzle tables for safe CRUD operations.

When to use it

• Creating or updating table definitions under src/database/schemas/
• Designing many-to-many relationships (junction tables)
• Adding timestamps, indexes, or foreign keys with cascades
• Generating and reviewing SQL migrations for Postgres
• Inferring TypeScript types from Drizzle tables for insert/select operations

Best practices

• Use plural snake_case for table names and snake_case for columns
• Prefix IDs to indicate entity type; use uuid for internal tables
• Spread standard timestamps from helper functions (createdAt, updatedAt, accessedAt)
• Return arrays for indexes (object style is deprecated); prefer uniqueIndex named explicitly
• Make migrations idempotent (IF NOT EXISTS / DROP IF EXISTS) and name files meaningfully

Example use cases

• Define agents table with id text primary key, user reference, config jsonb, and timestamps
• Create a agents_knowledge_bases junction table with composite primary key and enabled flag
• Add a new nullable column to users using an idempotent ALTER TABLE ... IF NOT EXISTS migration
• Generate TypeScript types with createInsertSchema and use $inferInsert / $inferSelect in services
• Add a unique index over client_id and user_id using the (t) => [...] index pattern