hermes - 💡(How to fix) Fix [IMPROVEMENT] `no_agent=True` cron jobs silently produce no output on failure [1 participants]

Problem: When a cron job is created with no_agent=True, it runs as a pure script watchdog — raw stdout is delivered verbatim, no token usage is logged, no agent trace exists. A script that exits 0 without producing output (e.g., a condition was never met, an exception was caught silently) will silently do nothing. The user receives no alert and the failure goes undetected.

1. Non-zero exit = always alert — no_agent=True scripts with non-zero exit should always trigger an error alert, regardless of stdout

TL;DR

Implementing a non-zero exit alert and heartbeat interval for no_agent=True cron jobs can help detect silent failures.

Guidance

• Consider implementing the proposed fix of triggering an error alert for no_agent=True scripts with non-zero exit, regardless of stdout.
• Add a heartbeat_interval to no_agent=True jobs to ensure they emit a signal within the specified time or trigger an alert.
• Modify the hermes cron list command to display no_agent jobs distinctly and show their last run status and exit code for better visibility.
• Review the current cron job configuration and script behavior to identify potential silent failure scenarios.

Example

No code snippet is provided as the issue does not contain specific code details.

Notes

The proposed fixes aim to address the issue of silent failures in no_agent=True cron jobs, but the implementation details may vary depending on the specific use case and system configuration.

Recommendation

Apply the proposed workaround by implementing non-zero exit alerts and heartbeat intervals for no_agent=True cron jobs, as this will help detect and prevent silent failures.