What problem does it solve?
When a production release introduces critical bugs or outages, manually rolling back deployments across multiple cloud platforms (Vercel, Fly.io, AWS Amplify) is slow, error-prone, and risks accidental rollbacks or partial failures that leave services in an inconsistent state.
Core Features & Use Cases
- Mandatory Typed Confirmation: Requires operators to explicitly type "ROLLBACK PROD" before any rollback actions run, eliminating accidental triggers during incident response.
- Configurable Multi-Target Rollback: Supports rolling back deployments in a custom order (e.g., backend services first, then frontend) via a simple deploy targets configuration file.
- Partial Failure Resilience: If one deploy target fails during rollback, the skill reports the partial state clearly and does not auto-revert targets that already rolled back successfully, giving operators full visibility into the current system state.
Use case: If a recent Vercel frontend deploy breaks user authentication while your Fly.io backend remains stable, you can safely roll back just the frontend or both targets in the correct order with full auditability of what succeeded and what failed.
Quick Start
Use the rollback-prod skill to safely revert your most recent production release across all configured deploy targets after confirming the action when prompted.
