Rollback Designer

Rollback Designer

Goal

Make risky action reversible enough to proceed safely.

This skill does not block change by default. It defines how to back out, contain damage, or fail safely.

When To Use

• before risky changes
• before rollouts, migrations, or broad edits
• when failure cost is meaningful
• when "we can undo it" is being assumed, not designed

Scope Boundaries

In scope:

• define rollback triggers
• define safe fallback states
• define what must be backed up or preserved
• name the smallest safe rollout pattern

Out of scope by default:

• full incident response plans
• exhaustive reliability engineering

Workflow

1. State the risky action.
2. Name the failure modes that would force retreat.
3. Define the last known safe state.
4. Identify what must be preserved for rollback.
5. Define rollback trigger, owner, and order.
6. Recommend the safest rollout shape.

Output Contract

Always return:

1. risky_action
2. failure_triggers
3. safe_state
4. rollback_requirements
5. rollback_path
6. recommended_rollout_shape
7. residual_risk
8. recommended_followup
9. next_step

Guardrails

• Do not assume rollback is trivial.
• Do not proceed without naming the safe state.
• Prefer staged rollout over all-at-once when possible.

Anti-Patterns

• imaginary rollback
• no trigger for stopping
• irreversible change with no containment