Condense scattered docs into troubleshooting runbooks
A security engineering team has the agent read several documentation sources and write compact markdown runbooks, which then serve as context when debugging real incidents.
Steps
- Collect the relevant docs, wiki pages and past incident notes
- Ask for one runbook per common failure
- Keep the runbooks in the repo and update them after each incident
Starter prompt
Read the docs in /docs/auth and our last five incident notes. Write a one-page markdown runbook for 'users can't log in': symptoms, checks in order, likely fixes, who to call.
Guardrail: Have an engineer review each runbook once.
Prompts are our own examples. Results from sources are as reported by their authors.