Skip to content
DevOps AI ToolKit
Newsletter
All prompts
Reduce MTTR with AI Difficulty: Intermediate ClaudeChatGPTCursor

On-Call Readiness & Paging-Coverage Audit Prompt

Audit whether a page will actually reach an awake, empowered human fast — rotation gaps, missing fallbacks, stale contacts, unacked escalation — so time-to-acknowledge doesn't silently blow up the front of every incident's MTTR.

Target user
SRE leads and on-call program owners
Difficulty
Intermediate
Tools
Claude, ChatGPT, Cursor

The prompt

You are an on-call program reviewer. The goal is to find every way a page could fail to reach an awake, empowered human quickly — because time-to-acknowledge is dead time at the front of MTTR that no amount of fast diagnosis can recover.

Give me:
- The rotation setup: [SCHEDULES, TIMEZONES, PRIMARY/SECONDARY, HANDOFF TIMES]
- Escalation policy: [WHO GETS PAGED, AFTER HOW LONG UNACKED, HOW MANY LEVELS]
- Contact and delivery: [PAGE CHANNELS — push/SMS/call — AND HOW CONTACTS ARE KEPT CURRENT]
- Coverage constraints: [TEAM SIZE, FOLLOW-THE-SUN OR NOT, KNOWN THIN SPOTS]

Work through this:

1. **Find coverage gaps.** Identify any window — timezone seams, weekends, holidays, handoff overlaps — where no one is clearly primary, or where a single person is the only coverage. List each gap and the risk it creates.

2. **Test the escalation chain on paper.** Trace what happens when the primary doesn't ack: how long until secondary, until a manager, until someone with authority to make big calls. Flag any chain that dead-ends, loops, or takes too long to reach a decision-maker.

3. **Check delivery reliability.** Assess whether the page can actually wake someone — redundant channels, do-not-disturb bypass, tested devices — versus relying on a single push notification that a sleeping phone swallows.

4. **Check authority, not just presence.** Confirm the person paged can actually act (access, permissions, decision authority). Being reachable but unable to mitigate is a hidden acknowledgement-to-action gap.

5. **Prioritize fixes.** Rank the gaps by how much acknowledgement delay they could cause and how easy they are to close.

Output format: a "READINESS AUDIT" — a table of GAP, WHEN IT BITES, ACK DELAY RISK, FIX, EFFORT — plus a short prioritized remediation list. Base findings only on the provided config; where you're inferring, say so. Recommend fixes; do not send test pages or change schedules yourself.

Run this prompt with AI

Test it, get an AI-improved version, or compare models — live in the Prompt Workspace. No copy-paste.

Why this prompt works

Every MTTR breakdown implicitly assumes the clock starts when a human engages — but the front of that clock, time-to-acknowledge, quietly balloons when the paging system itself has holes. A page that lands in a timezone seam, escalates into a dead-end chain, gets swallowed by a sleeping phone’s do-not-disturb, or reaches someone who lacks the access to actually act, costs minutes or tens of minutes before diagnosis even begins. And because these gaps only bite during real incidents at 3 a.m., they stay invisible in normal operation.

This prompt makes that invisible surface auditable on paper. It hunts specifically for the structural failure modes — uncovered windows, single-person coverage, escalation chains that loop or stall, non-redundant delivery — and adds the one teams routinely miss: authority. Someone who is reachable but cannot access production or make the call is an acknowledgement-to-action gap dressed up as coverage. Naming each gap with when it bites and its ack-delay risk turns a fuzzy worry into a prioritized fix list.

The guardrails keep the audit from becoming its own incident. It works from configuration rather than firing surprise test-pages that erode pager trust, and it defers staffing and escalation changes to the humans affected — because a plan that looks like full coverage but silently depends on one person always answering is exactly the failure it is meant to catch. Closing these gaps compresses the most frustrating part of MTTR: the dead time before anyone has even started.

Related prompts

More Reduce MTTR with AI prompts & error guides

Browse every Reduce MTTR with AI prompt and troubleshooting guide in one place.

Free download · 368-page PDF

Reading prompts? Get all 500 in one free PDF

500 battle-tested, copy-paste AI prompts engineered by a senior systems engineer — every one with fill-in placeholders and safety/back-out notes. Drop your email and it's yours.

  • 500 prompts: Linux · Kubernetes · Terraform · OpenStack · GitLab · Docker · Monitoring · Incident Response
  • Instant PDF download — yours free, forever
  • Plus one practical AI-workflow email a week (no spam)

Single opt-in · unsubscribe anytime · no spam.