Do not wrap the client in your own retry loop without knowing the default: transient 429s and 5xx are already retried twice. Tune maxRetries up for batch jobs or down when you want fast failure. Catch 4xx errors (non-retried) separately since they indicate a real request problem.

Context: Official docs (Reducto Node SDK README): documents that the client retries connection errors plus 408, 409, 429, and 5xx responses twice by default, with maxRetries configurable.