incident-postmortem-writer
Turns incident notes and timelines into a blameless postmortem with impact, root cause, timeline, what went well and action items. Use when the user needs to write up an outage, a failed deploy, a data problem or a security incident.
#writing #sre
Install
npx skills add https://github.com/Lethe044/agentskill-kit/tree/main/skills/incident-postmortem-writer
claude plugin marketplace add https://llmmart.ai/marketplace.json && claude plugin install lethe044-agentskill-kit@llmmart
git clone https://github.com/Lethe044/agentskill-kit.git
The skills CLI installs just this skill, for any of its supported agents. Claude Code installs the whole lethe044/agentskill-kit collection as a plugin from our marketplace. Git is the plain clone.
Skill manifest
Incident postmortem writer
A postmortem explains what happened and how to make it less likely, without blaming a person.
When to use
- The user pastes chat logs, alerts or notes from an incident and wants a write-up.
- A review meeting is coming and the document is not started.
- The user asks how to structure a postmortem.
Format
# Postmortem: <short title>
**Date:** YYYY-MM-DD **Duration:** Xh Ym **Severity:** SEV-N **Status:** draft
## Summary
Two or three sentences: what broke, who was affected, how it was fixed.
## Impact
Users or systems affected, for how long, with numbers when known (requests failed, orders lost).
## Timeline (UTC)
| Time | Event |
| --- | --- |
| 14:02 | Alert fired for elevated error rate |
## Root cause
The technical cause, then the contributing factors that let it happen or delayed detection.
## Detection and response
How it was noticed, how long it took, what slowed the response.
## What went well
## What went wrong
## Where we got lucky
## Action items
| Action | Owner | Due | Type (prevent, detect, mitigate) |
| --- | --- | --- | --- |
Rules
- Blameless. Describe actions and systems, not people. Write "the deploy script did not check the schema", not "Sam forgot to check".
- Facts before conclusions. Build the timeline first from the sources. Mark gaps as unknown.
- Never invent. Do not make up times, numbers, owners or causes. Write
unknownorto be confirmed. - Find the system cause. Ask "why" until the answer is a missing check, a gap in monitoring or a weak default, not a human mistake.
- Action items must be checkable. Each has an owner and a date, and fixes one cause. Prefer few strong items over a long list.
- Keep secrets out. Remove tokens, customer data and internal hostnames that do not need to be shared.
Errors
If the notes have no times, produce the format with the timeline marked "to be filled in" and list the questions that would fill it.
Files (agentskill-kit)
-
SKILL.md 2.4 KB
--- name: incident-postmortem-writer description: Turns incident notes and timelines into a blameless postmortem with impact, root cause, timeline, what went well and action items. Use when the user needs to write up an outage, a failed deploy, a data problem or a security incident. license: MIT metadata: version: 1.0.0 author: Lethe044 tags: [incident, postmortem, sre, writing] --- # Incident postmortem writer A postmortem explains what happened and how to make it less likely, without blaming a person. ## When to use - The user pastes chat logs, alerts or notes from an incident and wants a write-up. - A review meeting is coming and the document is not started. - The user asks how to structure a postmortem. ## Format ```markdown # Postmortem: <short title> **Date:** YYYY-MM-DD **Duration:** Xh Ym **Severity:** SEV-N **Status:** draft ## Summary Two or three sentences: what broke, who was affected, how it was fixed. ## Impact Users or systems affected, for how long, with numbers when known (requests failed, orders lost). ## Timeline (UTC) | Time | Event | | --- | --- | | 14:02 | Alert fired for elevated error rate | ## Root cause The technical cause, then the contributing factors that let it happen or delayed detection. ## Detection and response How it was noticed, how long it took, what slowed the response. ## What went well ## What went wrong ## Where we got lucky ## Action items | Action | Owner | Due | Type (prevent, detect, mitigate) | | --- | --- | --- | --- | ``` ## Rules 1. **Blameless.** Describe actions and systems, not people. Write "the deploy script did not check the schema", not "Sam forgot to check". 2. **Facts before conclusions.** Build the timeline first from the sources. Mark gaps as unknown. 3. **Never invent.** Do not make up times, numbers, owners or causes. Write `unknown` or `to be confirmed`. 4. **Find the system cause.** Ask "why" until the answer is a missing check, a gap in monitoring or a weak default, not a human mistake. 5. **Action items must be checkable.** Each has an owner and a date, and fixes one cause. Prefer few strong items over a long list. 6. **Keep secrets out.** Remove tokens, customer data and internal hostnames that do not need to be shared. ## Errors If the notes have no times, produce the format with the timeline marked "to be filled in" and list the questions that would fill it.
Comments (0)
Sign in to join the conversation.
Reviews (0)
No reviews yet.
No comments yet.