Bulletproof Make Scenarios: Error Handling Patterns That Actually Work

Design for failure first

Every external call is a coin flip on a bad day. Add a break route with an exponential retry ladder to every HTTP, database and third-party module before the scenario ever reaches production.

Dead-letter, do not delete

Route unrecoverable executions to a dedicated data store instead of dropping them. A weekly triage of that store is cheaper than the customer emails you would otherwise receive.

Idempotency keys everywhere

Any scenario that writes to a downstream system needs a deterministic key derived from the source event. Reruns then become safe by construction, which is what makes aggressive retries acceptable.