Repair Database Migration Rollback Bugs

The problem

Database migrations often fail during rollback or cause downtime because dual-write windows, nullable transitions, and application compatibility are handled inconsistently across the repository. Manual fixes frequently overlook state ownership or cleanup logic, leading to data corruption when a deployment is reverted.

What it does

• Identifies defects in expand, migrate, and contract phases to ensure zero-downtime deployments.
• Audits old and new application code to maintain compatibility during rolling updates.
• Fixes dual-read/write window logic and backfill ownership bugs.
• Validates index and constraint rollout strategies against rollback boundaries.
• Generates deterministic local fixtures and regression tests for migration verification.

Why this beats prompting it yourself

General prompts often ignore the "contract" phase of a migration or fail to account for concurrent state transitions. This skill enforces a canonical contract for identity and validation, ensuring that side effects and retries are bound to the specific state that created them rather than relying on broad retries or sleeps.

Use cases

• Fixing a migration that prevents an application from rolling back to a previous version.
• Implementing safe dual-write logic for a high-traffic table refactor.
• Correcting inconsistent nullability or default value transitions across microservices.
• Adding observability and deterministic tests to brittle migration runbooks.

Known limitations

Does not interact with production databases, live secrets, or deployed infrastructure. Requires local repository files and fixtures to perform analysis and patching.