The recurring hygiene sweep for tickets going quiet. Distinguish tickets genuinely stalled on the
MSP's side from tickets correctly waiting on a client, vendor, or scheduled date — and produce a
prioritized action list, not a raw age report.
1. Enumerate open tickets, splitting searches per board and per status so result caps don't
silently truncate the sweep. If any search may have capped, say so.
2. For each ticket, compute the last meaningful client-facing update: the most recent
client-visible message or substantive progress note. Internal notes, time entries, status
flips, and automated touches do NOT reset the clock — a ticket "touched" daily by a bot can
still be stale.
3. Classify each stale ticket's ownership:
- MSP-owned stall — the ball is in our court: client replied and we went quiet, work promised
but not documented, unassigned and aging.
- Legitimate wait — client response pending WITH a follow-up actually sent, vendor RMA or
third-party dependency documented, or a future scheduled service date in a proper scheduled
status. Verify the wait is documented, not assumed.
4. For each MSP-owned stall, leave an internal plain-text note @mentioning the assigned tech —
look them up to resolve the mention — stating days since the last meaningful client-facing
update and the single next expected action. If unassigned, flag for dispatch instead.
@mention notes are factual nudges, not blame.
5. Pull tickets nearing or past an SLA target out of the routine list and flag them at the top —
imminent breaches are not routine aging.
6. For legitimate waits with no future touchpoint, offer to schedule a follow-up date rather than
leaving them adrift. Skip tickets parked on a future scheduled call in a proper status — don't
nag a correctly scheduled ticket.
7. Output a prioritized list: SLA-imminent first, then MSP-owned stalls (oldest meaningful-update
first), then legitimate waits with their next touchpoint. One recommended action per ticket:
nudge, escalate, schedule, or route to closure.
8. Present everything for sign-off before acting. Never bulk-close, bulk-reassign, or bulk-send
follow-ups without explicit confirmation — closure of no-response tickets belongs to the
No-Response Closure Sequence skill after documented attempts. Disclose result caps rather than
presenting a truncated sweep as complete. If write tools are disabled, deliver the action list
in chat only.
