Build and release code has a different discipline than feature code: it must be reproducible, abortable, and reversible. A build that depends on whatever the last run left behind is unreliable by definition. Before you change a build step, ask what happens if it crashes halfway, and whether you could undo the release after it ships.

Fires when

Editing a build script, a release or packaging step, a publish flow, or CI config.

How to apply

Start from a guaranteed clean state: remove stale artifacts and intermediates, restore any patched config, before you begin.
Smoke-test the smallest case first and abort fast if it fails; don't run the full matrix on a setup you haven't proven.
Version artifacts into timestamped outputs instead of overwriting, so rollback is trivial.
Validate outputs, not just inputs. A silently-empty artifact, or a "0 tests passed" nobody asserted on, is the worst failure mode and the easiest to miss.
Protect any global state you patch-and-restore with a lock, so a half-finished run doesn't leave the machine broken.
Know the rollback before you ship.

Worked example

A release script builds against eight SDK versions over an hour, then you discover the first one was broken from the start — the whole hour is wasted. A canary fixes it: build the smallest target first, and abort the run if it fails, so a broken setup costs two minutes instead of sixty. The same instinct catches a "0 tests passed" that would otherwise ship green because nobody asserted the count was above zero — a loud failure is always better than a silent bad artifact.

Red flags

Thought	Reality
"The machine's probably clean"	Probably is not a guarantee; start clean.
"It built, so it's good"	Built is not the same as validated. Check the output.
"I'll just overwrite the last build"	Then you can't roll back when this one's bad.
"All eight will be fine"	Smoke-test one before you spend an hour on eight.