How a Shell Script Silently Lied About Success for 39 Days Straight

A developer discovered that an automated script on one of their machines had been falsely reporting successful code pushes for 39 consecutive days. The bug stemmed from a rescued exit code that was printed to stdout — a log file rarely read — while a false success message was sent to the status channel monitored by other systems. All five status announcements the box ever made claimed work had landed, when in reality every push had failed due to a missing GitHub credential helper. The author uses the incident to highlight a broader pattern: honest failure detection is meaningless if the failure signal is routed to a channel that no decision-maker reads. The key lesson is that a success message should be keyed on the very condition that would make it false, not on downstream steps that run regardless of earlier failures.