Single Shell Script Can Detect Silent Failures in Unattended Claude Automation
A developer has shared a shell script called automation-health.sh designed to monitor unattended AI automation jobs running via macOS launchd. The script addresses the problem of silent failures, where crashed or broken jobs show no outward signs of malfunction — one such failure reportedly went unnoticed for six days. It checks multiple components at once, including launchd job statuses, hook script permissions, and log freshness, returning a single verdict of GREEN, WARN, or FAIL. Each check uses distinct exit codes, allowing the script to be chained into daily briefings or CI pipelines for automated follow-up. The approach eliminates the need to manually inspect each automation component individually, providing a consolidated health overview in one command.
This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)
Log in to join the discussion and vote.
Log in