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

Incident Postmortem Action Tracking Agent

Every postmortem action completed and verified

Incident Postmortem Action Tracking Agent: what goes in, what the agent does and what you get

What it does

After an incident the team agrees on good actions, but they often fade, and the same incident happens again. This agent reads the postmortem, extracts each action with an owner and a due date, and creates tickets. It asks for missing owners or dates before filing anything. Each week it checks ticket status and, where possible, verifies completion with evidence, such as a new alert rule existing or a config change applied. If a ticket is closed but the evidence is missing, it reopens it and tells the owner what is missing. It sends a weekly summary of overdue items. You approve escalations to managers. Edge case: an action that turns out to be no longer needed is closed only with a written reason.

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
Yes, continueYes, continueApprovedNoNo 1 STARTS WHEN Postmortem published 2 DOES Extract actions, owners and due dates 3 CHECKS THE RESULT Does every action have an owner and due date? If not: ask the incident lead to fill them in. Back tostep 2. 4 USES A TOOL Create tickets for each action 5 USES A TOOL Weekly check of ticket status and evidence 6 CHECKS THE RESULT Is each closed action backed by evidence? If not: reopen the ticket and ask for proof. Back tostep 5. 7 YOU APPROVE Lead approves escalation of overdue items 8 RESULT All actions verified and postmortem closed
Read the steps as a list
  1. Postmortem published
  2. Extract actions, owners and due dates
  3. Does every action have an owner and due date?If not: ask the incident lead to fill them in. Back to step 2.
  4. Create tickets for each action
  5. Weekly check of ticket status and evidence
  6. Is each closed action backed by evidence?If not: reopen the ticket and ask for proof. Back to step 5.
  7. Lead approves escalation of overdue itemsThe agent waits here for your OK.
  8. All actions verified and postmortem closed

How it decides

An action is done only when its ticket is closed and its evidence is found. Actions overdue by the set days are escalated.

  • Escalate actions overdue by more than 14 days
  • Close unneeded actions only with a written reason
  • Link repeat incidents to earlier open actions

Make it yours

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

  • Overdue days before escalation (default 14)
  • Evidence checks per action type
  • Summary day and recipients
  • Ticket project to use

What keeps you in control

It always asks you first

  • Escalating to senior managers
  • Closing a postmortem

Hard limits

  • Read-only access to systems used for evidence
  • No escalation without approval

It stops when

  • Done: every action verified or closed with reason
  • Stop: postmortem has no actions

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 breach review at Harborview Logistics on February 2 listed 9 actions. Two had no owner, so the agent asked the incident lead before filing. Four weeks later, the action to alert on admin logins from new countries was closed, but the evidence check found no such rule. The agent reopened it. The rule went live a week later, and the security lead approved closing the review.

More agents for site reliability engineers