← Changelog
FixedAuditor

An automation or Issue assistant run that fails now says so

A run started from a GitHub issue or pull request that failed, ran out of time, or was stopped by the balance or a budget used to leave a reaction and then nothing, and a scheduled automation could fail every week with no digest sent. Now the run comments on the issue or pull request with the reason and what to do, the email and Slack digests go out for failed firings with the failure in the subject, and a new webhook event, auditor.automation.failed, carries the run and its reason.

A run that the balance, a budget or the team's Directives cut short used to be recorded as done with nothing to do. It now ends with its own status, stopped, on the run's page, in the API and in the digest, and keeps its conversation: asking again after topping up continues where it stopped.

To keep pull requests quiet, a run nobody asked for says the same thing once per issue or pull request in a week, and says nothing when the automation was turned off or the pull request closed before it ran.

More in When a run doesn't finish.