Copy-paste the block below into any AI coding agent to review a database or data model change: normalization, indexes, migrations, and integrity rules, each backed by a file:line or a query result.
Keywords: data-model, schema, database, migration, normalization, index, review
Review `[the schema change, migration file, or table]` in this repository. Work from evidence: apply the migration, read the queries, and cite what you actually ran. ## Steps 1. **Read the change** - Read the migration files, model definitions, and the queries that touch them. Do not review from the diff summary alone. 2. **Check normalization** - Confirm entities and relationships match the requirements, and name any redundancy with the reason to keep it. 3. **Verify indexes against real reads** - Find the actual query patterns and confirm an index supports each. Flag both missing indexes and redundant ones, with the query that justifies the finding. 4. **Review migration safety** - Confirm backward compatibility during rollout, a rollback path, and that existing rows transform without loss. 5. **Check integrity rules** - Confirm foreign keys, constraints, and validation exist where the requirement needs them, not just where the ORM generated them. 6. **Report with evidence** - Every finding cites a file:line or a query result. Separate blocking issues from optional improvements. ## Verification - [ ] The migration was applied and rolled back against a real database. - [ ] Every finding cites a file:line or a query result. - [ ] Each index recommendation is tied to a query that needs it. - [ ] The rollback path is stated and was tested, not assumed. - [ ] Blocking issues and optional improvements are listed separately. ## Rules - Never approve a schema change without migration and rollback evidence. - Never recommend an index without the read pattern it serves. - If the requirement does not justify the normalization, name the part that is missing. - If you could not run the migration, say so instead of implying it passed.