# Write AI troubleshooting prompts that produce reviewable answers

Educational worksheet; not an executable workflow.
Reference: https://netprompt.ai/learn/network-troubleshooting-ai-prompts

Task / service:
Scope and owner:
Platform and version:
Observation window and time zone:

## Practical example
Evidence review: “Using observations E1–E4, list confirmed facts, possible explanations and missing evidence. Reference the observation for each finding. Do not infer that an unavailable check passed.”
Comparison: “Compare the affected and healthy locations using the same time window. Identify meaningful differences and explain what additional check would distinguish the leading explanations.” Escalation: “Prepare a handoff containing impact, checks completed, unresolved questions and the next responsible team. Keep hypotheses labeled.”

## Review checklist
- [ ] Replace placeholders with verified service and platform details.
- [ ] Use stable labels for evidence and timestamps.
- [ ] State the intended output and coverage limits.
- [ ] Ask what could disprove the leading explanation.
- [ ] Review commands and actions outside the prompt itself.

## Evidence review
| Claim or question | Observation / source | Time | Known gap | Next action / owner |
|---|---|---|---|---|
| | | | | |

## Outcome
Supported findings:
Unresolved hypotheses:
Required approval, if any:
Verification result:
Reviewer and date:
