AGENT SKILL

incident-triage

Triage customer-reported service incidents when production impact, errors, authentication failures, or integrations are mentioned.

License 许可
Apache-2.0
所属仓库
Usefused/harnest
最近更新
2026-11-14
Stars
★ 3
Forks
1

01 前言 Front Matter

name
incident-triage
description
Triage customer-reported service incidents when production impact, errors, authentication failures, or integrations are mentioned.

03 SKILL.md 正文

Incident triage

Use this procedure when a customer reports a service incident. It produces a recommendation only; the host application, not the agent, owns ticket creation.

  1. Separate observed symptoms from assumptions. Never invent logs, causes, or recovery times.
  2. Determine whether the customer explicitly says production is blocked. Do not infer a production block merely from urgency or frustration.
  3. Call triage_request with a concise symptom summary and the confirmed production_blocked value.
  4. Report the returned queue and priority, then ask for the smallest useful missing diagnostic detail, such as an error code or request ID.
  5. Delegate root-cause investigation to technical_specialist when the request involves an API, integration, authentication, or reproducible error.

Do not request passwords, API keys, session cookies, or full authentication headers. Ask customers to redact credentials from diagnostic material.

If diagnostic material already includes credentials or personal information, omit those values from replies, triage_request arguments, and technical_specialist handoffs. Use only the minimum non-sensitive symptoms, error codes, and confirmed production impact needed for triage. Do not copy raw sensitive material into files, logs, tickets, or external services. Ask for a redacted excerpt when those facts cannot be separated safely; do not claim to erase material already supplied to the host application.

04 同仓库更多 Skills