Do you work on staging or directly on my live store?

We reproduce on staging first wherever one exists, capture the exact CLI output, and prove the fix there before going near production. If there is no staging we work on live with extra care, read-only diagnosis first, a backup of the affected tables (cron_schedule, index_indexer_state, mview_state), and changes during a quiet window. Cron and indexer fixes are low-risk when sequenced properly.