What problem does it solve? Changing legacy APIs, database schemas, configs, or feature flags without breaking existing consumers is risky: silent data loss, unbounded backfills, missing rollback paths, and premature deletion of old code are common failure modes. This Skill provides a structured methodology for planning and verifying migrations, deprecations, and rollouts so old and new behavior coexist safely until retirement is proven. ## Core Features & Use Cases - Migration Mapping: Identifies current state, target state, source of truth, consumers, compatibility window, rollout control, and rollback path before any edit. - Safe Patterns Library: Covers expand/migrate/contract, dual-read, dual-write, feature flags, and shadow mode with explicit failure and mismatch policies. - Data Migration Safety Rules: Enforces backups, idempotent bounded batches, progress checkpoints, count/checksum verification, and PII-safe logging. - Use Case: When replacing a username-based auth session model or migrating a PostgreSQL schema, use this Skill to produce a plan covering dual-read fallback, staged rollout behind a flag, dry-run verification, and a tested rollback path. ## Quick Start Ask the AI to plan a safe migration of the legacy API endpoint to the new schema, including a compatibility window, dual-read fallback, and a verified rollback path.
