What problem does it solve? Firestore migrations in this project run directly against production data for an active EMS corps, with no staging copy or point-in-time restore. This Skill enforces the discipline needed to write and run migration scripts, backfills, and data fixes without corrupting live inventory records. ## Core Features & Use Cases - Dry-run-first workflow: All migration npm scripts default to dry-run mode, requiring explicit flags and user approval before any live write executes. - Migration script standards: New scripts must be idempotent, log every change with before/after values, batch writes under the 500-op Firestore limit, and preserve audit trails in inventory_logs and auditEvents. - Emulator rehearsal: Changes are rehearsed against the local Firestore emulator with seeded legacy fixtures before any production dry-run is proposed. - Use Case: When renaming a field across inventory documents, use this Skill to write an idempotent migration script, rehearse it on the emulator, review the dry-run output, and only then execute the live write with validation checkers. ## Quick Start Ask the AI to write a dry-run-first Firestore migration script that backfills a missing field on inventory documents and rehearse it on the emulator before touching production.