Make your code safe to retry after a crash.
Guides the design of operations that reach the same correct state after retries, restarts, or interrupted runs.
When to use it
Use when designing commands, lifecycle steps, or processing loops that must handle crashes, restarts, and retries.
It changes how the agent designs state-mutating operations by requiring repeat-run and crash-recovery checks, with reconciliation when leftover state affects results.
What you provide
No setup requirements listed.
No additional actions listed.