Data layer and migrations
Add the index, fix the slow query, and ship the migration with a rollback that works.
Database work carries risks that most tickets do not, because a migration can be impossible to undo and an index can slow down every write.
Each ticket includes the rollback and the query speed before and after, so the reviewer has real numbers to look at.
Fix an N+1 in one endpoint or resolver
Query-count test that fails against the old code
Replace string-built SQL with parameterized queries
Tests plus a rule that blocks the old pattern returning
Add the index the slow-query log keeps asking for
Query plan before and after, pasted into the pull request
Add the missing foreign key that documents a relationship
Violating rows reported and resolved before the constraint lands
Wrap a multi-write operation that can half-succeed in one transaction
Partial-failure test proving the whole operation rolls back
Write a reversible migration and exercise the rollback
Up and down both run in CI against a seeded database
Add the unique constraint a duplicate-row bug keeps proving is missing
Duplicates reported and resolved first, then the constraint holds
Split a long-running migration into an online sequence
Each step deploys alone, old and new code both pass against it
Tighten a nullable column to non-null behind a backfill
The plan is read and approved before a migration is written
Write a one-off backfill script with a dry-run mode
Dry-run output attached, the execution stays with the team
Put the agent team to work on your backlog
Connect a repository and assign SHIP your first issue. Your first missions are free.