返回 Skill 列表
extension
分类: AI Agent 能力无需 API Key

incident-postmortem

基于证据的根本原因分析和系统性预防措施的事后工作流程。在事件稳定后使用,当团队需要时间线、根本原因链、促成因素以及负责的预防措施时;不要用于主动事件指挥和实时缓解。

person作者: jakexiaohubgithub

Incident Postmortem

Overview

Use this skill to turn incidents into concrete learning and prevention changes, not just narrative summaries.

Scope Boundaries

  • Use this skill when the task matches the trigger condition described in description.
  • Do not use this skill when the primary task falls outside this skill's domain.

Shared References

  • Root-cause evidence rules:
    • references/root-cause-evidence-rules.md

Templates And Assets

  • Postmortem report template:
    • assets/postmortem-report-template.md
  • Incident timeline template:
    • assets/incident-timeline-template.csv
  • Corrective action tracker:
    • assets/corrective-action-tracker-template.csv

Inputs To Gather

  • Incident timeline signals and operational evidence.
  • Customer/business impact and severity context.
  • Mitigation and recovery actions taken.
  • Known system/process constraints.

Deliverables

  • Evidence-backed postmortem report.
  • Timeline artifact linking events to signals.
  • Corrective action plan with owner, due date, and verification.
  • Residual risk statement and follow-up tracking.

Workflow

  1. Build timeline in assets/incident-timeline-template.csv.
  2. Analyze trigger, root causes, and contributors using references/root-cause-evidence-rules.md.
  3. Draft report in assets/postmortem-report-template.md.
  4. Track prevention actions in assets/corrective-action-tracker-template.csv.
  5. Publish and align owners on verification timelines.

Quality Standard

  • Root cause is evidence-backed, not speculative.
  • Timeline and impact statements are explicit and auditable.
  • Corrective actions are specific, owned, and verifiable.
  • System/process improvements are prioritized over blame narratives.

Failure Conditions

  • Stop when root-cause claims lack supporting evidence.
  • Stop when action items have no owner or verification method.
  • Escalate when residual risk exceeds accepted reliability policy.