| name | planning-rollback-mechanism |
| description | Rollback mechanics and scenario: criteria, steps, data impact, recovery |
Create a fast and safe path to revert in case of degradation
Change description + rollout strategy
Data/compatibility constraints (especially for migrations)
Available mechanisms: revert, feature flag, blue-green
Define rollback criteria (degradation signals, thresholds, time)
Select a rollback mechanism and estimate recovery time (RTO)
Describe rollback steps (operational actions) and order
Describe data impact and compatibility after rollback
Define post-rollback verification (at minimum: key scenarios + metrics)
<output_format>
Rollback mechanism + expected time
Data impact / compatibility
Post-rollback verification
<quality_rules>
Rollback is feasible within a clear timeframe and described step-by-step
Data impact is stated (what happens to partially applied changes)
</quality_rules>
<do_not>
Do not leave rollback as "we will roll back if needed" without mechanics
Do not plan a rollback that requires manual data edits without verification
</do_not>