CLAUDE CODE AGENTS: ADVANCED ORCHESTRATION - Chapter 4, Exercise 1
Choosing the Right Wake-Up Delay for a Six-Minute Deployment
====================================================================================

QUESTION:
An agent is waiting on a deployment pipeline that typically takes about six
minutes to finish. Explain what wake-up delay would be appropriate, and
why checking every 15 seconds would be the wrong choice.

SOLUTION / EXPLANATION:

This chapter is explicit that the right wake-up delay should match how fast
the thing actually being waited on realistically changes - in this case,
an external deployment pipeline you don't directly control, which typically
takes about six minutes. An appropriate delay is one check scheduled around
that same timeframe - roughly six minutes, or slightly less to allow one
follow-up check if it isn't quite finished yet - rather than a much shorter
interval.

Checking every 15 seconds would be the wrong choice for a concrete,
practical reason: a deployment that typically takes six minutes is
extremely unlikely to have meaningfully progressed, let alone finished,
within any given 15-second window early in that six minutes. Checking that
frequently means the overwhelming majority of checks report back "still not
done" - the exact same answer, over and over, dozens of times, before the
deployment actually finishes. Each of those checks carries a real cost
(whatever resources a check actually consumes) for effectively zero new
information gained, since nothing meaningful could plausibly have changed
in 15 seconds for a process on a six-minute timescale.

The correct approach, per this chapter's own polling guidance, is matching
the check interval to the actual expected timescale of what's being waited
on - a six-minute process deserves a check around six minutes, not dozens
of nearly-instantaneous re-checks that mostly just confirm nothing has
changed yet.

--------------------------------------------------------------------------
WHY THIS WORKS AS AN ANSWER:
It applies the chapter's own "match the delay to the actual timescale"
principle concretely to the six-minute example, and explains specifically
why a much shorter interval produces mostly wasted, uninformative checks
rather than vaguely saying it's "too frequent."