Complete AI Training
Sign inGet my AI kit

Your job's AI kit

Get your AI kit

Tell us who you are and what you do. We show you your kit right away and email you the link: skills, prompts, AI agents, MCP servers and courses for your job.

500+ jobs ready, and we make a kit for any other job. No payment needed to look.

Share

AI agent for site reliability engineers

Runbook Freshness Test Agent

Runbooks that are tested, current and dated

Runbook Freshness Test Agent: what goes in, what the agent does and what you get

What it does

Runbooks go stale and fail during an incident, exactly when they are needed. This agent walks through each runbook step in a test environment or with read-only checks. It notes which commands, screens or settings no longer match, then drafts corrections. After edits, it runs the runbook again to confirm the new steps work. It also records the last tested date on each page. The owner approves the updated runbook. Edge case: a step uses a console menu that was renamed, so the agent finds the new name and updates the step and screenshot note.

How it works

Follow the arrows from top to bottom. The orange dashed arrow is the loop: when a check fails, the agent goes back and tries again.

Start and resultWhat it doesA check on its own workWaits for your OKGoes back and retries
ApprovedYes, continueNo 1 STARTS WHEN Runbook test is due 2 USES A TOOL Load the runbook and the test environment 3 DOES Walk through each step with read-only checks or intest 4 DOES Record steps whose commands or screens no longermatch 5 DOES Draft corrections for each failing step 6 YOU APPROVE Owner reviews the drafted corrections 7 USES A TOOL Apply the corrections to the runbook draft 8 CHECKS THE RESULT Does every step now pass when run again? If not: revise the failing step and rerun the test. Backto step 3. 9 DOES Add the test date and the tester to the runbook 10 RESULT Updated runbook ready for the owner to publish
Read the steps as a list
  1. Runbook test is due
  2. Load the runbook and the test environment
  3. Walk through each step with read-only checks or in test
  4. Record steps whose commands or screens no longer match
  5. Draft corrections for each failing step
  6. Owner reviews the drafted correctionsThe agent waits here for your OK.
  7. Apply the corrections to the runbook draft
  8. Does every step now pass when run again?If not: revise the failing step and rerun the test. Back to step 3.
  9. Add the test date and the tester to the runbook
  10. Updated runbook ready for the owner to publish

How it decides

A step passes when its command or screen exists and gives the expected result. A runbook passes when all steps pass.

  • Use test or read-only checks only
  • Fail a step when its result differs from the expected one
  • Test runbooks for critical services first
  • Update the tested date only after a full pass

Make it yours

Every agent is a starting point. You choose these settings for your own situation.

  • Test frequency (default: quarterly)
  • Critical runbook list
  • Test environment
  • What counts as a pass

What keeps you in control

It always asks you first

  • Owner approves the corrected runbook before it is published

Hard limits

  • Never runs a destructive step
  • Never publishes a runbook without the owner

It stops when

  • Done: all steps pass and the date is recorded
  • Stop: no test environment or read-only access is available

Set it up

We guide you through the set-up, step by step

Members get the full set-up guide for this agent. No technical skills needed: you copy, paste and upload.

10 minto set it up in your AI
5 AIsChatGPT, Claude, Copilot, Gemini, Grok
  • One set of instructions to paste into your AI, with the clicks for ChatGPT, Claude, Microsoft 365 Copilot, Gemini and Grok
  • The agent then walks you through connecting your own data, one source at a time
  • A downloadable copy with the flow chart, the rules and the full guide
Get access to this agent

An example run

What happensA database failover runbook had 12 steps. The agent found step 5 used a command renamed in the last upgrade and step 9 pointed to a removed dashboard. It drafted fixes. On the rerun, step 5 passed but step 9's new dashboard link needed a different role, so it added that. The owner approved all 12 steps tested.

More agents for site reliability engineers