Managing DB Migrations in casedotdev-mono

Quick checklist for committing schema changes with migrations in casedotdev-mono

Database Migration Checklist

Quick reference for schema changes. For full context, see managing-database skill.

Before Committing

# 1. Edit schema (packages/database/src/schema*.ts)

# 2. Generate migration (from repo root - REQUIRED)
bun db:generate

# 3. Review the latest generated migration
cat "$(ls -1 infra/drizzle/*.sql | sort -V | tail -1)"

# 4. Commit BOTH together
git add packages/database/src/schema*.ts infra/drizzle/
git commit -m "CD-XXX: Description (with migration)"

After Merge

Migrations are applied automatically by CI via bun db:migrate. No manual push step is needed.

  • Preview deploys run db:migrate against the staging branch
  • Production deploys run db:migrate against the production branch

Note: bun db:push now errors out intentionally. Use bun db:push:local only for throwaway local DBs.

Verification

Use Neon MCP to verify the migration was applied:

# Test enum value exists
neon_run_sql({ projectId: "billowing-paper-24157726", sql: "SELECT 'value'::enum_name;" })

# Test column exists
neon_describe_table_schema({ projectId: "billowing-paper-24157726", tableName: "table_name" })

CI Will Block If

  • Schema changed but no migration file committed
  • Migration file exists but _journal.json not updated
  • Schema and migration are out of sync (bun db:check fails)

Incident Reference: TG-45 (January 2026)

Multiple PRs merged with schema changes but no migration files. CI caught it at preview → main merge. Had to go back and generate migrations after the fact.

Don't be that engineer. Always run bun db:generate before committing schema changes.

Danger Zone