database-migrator_skill

Overview

This skill is a database schema migration and data transformation expert implemented for Rust-based agents. It provides practical guidance, safe migration patterns, and step-by-step processes to design, test, deploy, and roll back schema and data changes with minimal service disruption. Its focus is on zero-downtime, backward compatibility, and verifiable data integrity.

How this skill works

The skill inspects the current schema and identifies potential breaking changes, then generates a phased migration plan including scripts and rollback procedures. It recommends multi-step, non-blocking operations (add-then-backfill, create indexes concurrently/online) and prescribes testing and monitoring steps to validate integrity and performance. It supports common relational databases and outputs concrete actions for planning, development, testing, and deployment.

When to use it

• When changing column types, names, or nullable constraints on production tables
• When adding large columns or indexes to high-traffic tables
• When deploying schema changes across replicated or sharded environments
• When introducing data transformations or backfills that require staged rollout
• When you need a rollback-safe plan for irreversible changes

Best practices

• Design migrations to be backward compatible so old code continues to work during rollout
• Use multi-step changes: add new objects, backfill data, switch readers, then drop old objects
• Prefer online/lock-free operations (CONCURRENTLY, ONLINE DDL) to avoid long locks
• Always create backups and test rollback procedures on staging before production
• Monitor query latency, locks, replication lag, and error rates during migration

Example use cases

• Add a large TEXT column: add with default or nullable, backfill in batches, then set NOT NULL
• Change email column type: add email_new, backfill, switch application to email_new, drop old column
• Create an index on a live table using CREATE INDEX CONCURRENTLY or MySQL ONLINE DDL
• Perform a cross-table data transformation with scripted batch updates and integrity checks
• Roll out constraint changes by first enforcing them in application logic, then in DB