Check the health of your automations

One verdict per active scheduled or event task you can see, from its real runs in the window: delete (its last 5 runs all failed or went undelivered), consolidate (same instructions and trigger as an earlier task, named in duplicate_of), add_condition (runs more than 6 times a day and at least 70% of runs find nothing), run_less_often (runs more than daily and at least half find nothing), switch_model (a quality-tier task that mostly finds nothing), or keep. Each carries the monthly credits it is projected to spend, what the verdict would save, and how many of its latest runs failed in a row with the last error, plus the integrations that need reconnecting and their reconnect links. Read-only: nothing is changed. Report the verdicts and let the owner decide.