oncall-handofflisted

# On-call handoff A handoff passes the pager from the responder going off shift to the one coming on, along with everything the incoming person needs to not start cold. It fails when it becomes a dropped link and a "you've got it": the next engineer inherits open incidents with no thread, misses the deploy that is about to page, and rediscovers by outage the workaround that was never written down. The whole point is to move understanding, not just the pager. ## Method 1. **Do it live, with overlap.** Hold a short synchronous call, ten to fifteen minutes, where the outgoing responder walks the incoming one through the state. A wall of text nobody reads before their first page is not a handoff; a conversation where questions get asked is. 2. **Walk every open incident.** For each one give current status, what has been tried, who else is involved, the next planned step, and the ticket. An incident handed over as "still investigating" with no linked thread is one the next person restarts from zero at the worst time. 3. **List the watch items.** Name what is not paging yet but might: a deploy that went out an hour ago, a degraded upstream dependency, a certificate near expiry, a batch job that occasionally wedges. The next page usually comes from this list, so it deserves as much care as the active incidents. 4. **Transfer the tribal context runbooks do not hold.** The alert that is flaky and safe to ignore, the dashboard that lies, the service own