What problem does it solve? Database schema changes in a Drizzle/PostgreSQL project often ship with risky migrations: non-idempotent SQL that fails on re-run, blocking index creation on large tables, backfills crammed into deploy-time migrations, and rebase conflicts that corrupt migration journals. This Skill provides the conventions and rollout strategies to generate, review, and ship migrations safely. ## Core Features & Use Cases - Rollout strategy classification: Routes every schema change into one of three paths — regular Drizzle migration, online index creation with CONCURRENTLY, or a dedicated idempotent backfill script — validated against real Dev database measurements. - Migration hardening: Renames generated files meaningfully, adds idempotent clauses (IF NOT EXISTS, drop-then-add constraints), and keeps _journal.json tags in sync. - Development-stage cleanup: Regenerates draft migrations after schema churn, consolidates stacked branch migrations, and resolves rebase conflicts by regenerating from the rebased schema. - Use Case: When adding an index to a large production table, run CREATE INDEX CONCURRENTLY manually before deployment while keeping an idempotent non-concurrent version in the migration so deploys never block. ## Quick Start Ask the AI to generate a Drizzle migration for your schema change and review it for idempotency, rollout safety, and correct journal entries.
