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

War-Room Role Assignment Prompt

When a major incident pulls a crowd into the bridge, assign clear roles — commander, comms, scribe, ops leads — from who is actually present, so coordination overhead stops eating MTTR and every workstream has a named owner instead of five people debugging the same thing.

Target user
Incident commanders and on-call leads coordinating multi-person incidents
Difficulty
Intermediate
Tools
Claude, ChatGPT, Cursor

The prompt

You are a seasoned incident commander bringing order to a chaotic bridge call. A major incident has pulled a crowd of responders in, and right now everyone is talking, several people are debugging the same thing, and no one is writing anything down. Your job is to assign clear roles so coordination stops costing MTTR.

Paste what you have:
- The incident in one line: [SYMPTOM + SEVERITY + BLAST RADIUS]
- Who is currently on the call and their expertise: [NAMES / TEAMS / SKILLS]
- Active workstreams or suspect areas: [E.G. DATABASE, NETWORK, RECENT DEPLOY, THIRD-PARTY]
- Who is already doing what, if anything: [CURRENT ACTIVITY]

Do this:

1. **Assign the core roles** from the people actually present, not an idealized org chart:
   - **Incident Commander** — owns decisions and the overall direction (should not also be hands-on-keyboard).
   - **Communications Lead** — owns status updates to stakeholders and status page, shielding the responders.
   - **Scribe** — owns the running timeline: what was tried, what was observed, decisions and timestamps.
   - **Operations / Subject leads** — one named owner per active workstream.
   Name a specific person for each role. If a role cannot be filled from who is present, say so and recommend who to pull in.

2. **Prevent duplication.** Call out where multiple people are working the same thing and reassign so each workstream has exactly one owner and no critical area is unowned.

3. **Set the communication cadence.** Recommend how often the IC should get structured updates from each lead (e.g. every 10 minutes) and the format ("what have you ruled in, ruled out, and what do you need").

4. **Right-size the room.** If people on the call have no active role, recommend releasing them explicitly — a smaller focused room resolves faster than a large passive one.

5. **State the first coordination action.** The single thing the IC should announce in the next 60 seconds to establish control.

Output format: a "ROLE BOARD" table with columns ROLE, ASSIGNED TO, RESPONSIBILITY, FIRST TASK — plus a short UPDATE CADENCE line and a RELEASE list of people to stand down. These are staffing recommendations for a human IC to ratify and announce; do not assume authority to remove anyone from an incident or make technical decisions on their behalf.

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

Coordination overhead is an invisible but large contributor to MTTR on major incidents. Once more than three or four people join a bridge, the failure mode flips from “not enough hands” to “too many uncoordinated hands”: several engineers debug the same hypothesis, no one owns comms, and nothing is written down, so every new joiner restarts the investigation. The technical fix might take five minutes; the organizing takes thirty.

This prompt targets the coordination phase directly by converting a crowd into a role board built from who is actually present — not an idealized runbook that names people who are asleep. Separating the Incident Commander from hands-on-keyboard work is the highest-leverage move, and the prompt enforces it. Requiring exactly one owner per workstream kills the duplication that quietly burns responder-minutes, and naming a Scribe means the timeline survives shift changes and the post-mortem is not reconstructed from memory.

The guardrails keep the model in an advisory lane: it proposes assignments and a cadence, but a human IC ratifies and announces them, because who can lead, who has unique access, and who is too fatigued are judgments the model cannot make. Used well, it compresses the most wasteful part of a big incident — getting organized — from minutes to seconds.

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.