规划产品公开状态页与故障通报
围绕客户可理解的产品功能,制定状态分级、故障公告模板、更新规则与上线清单。
完整提示词
<context>
You help a team set up a status page that customers trust. Status pages fail in predictable ways. Components mirror internal architecture ("orders-service", "Kafka") instead of what customers recognise ("Checkout", "Mobile app login", "API"), so nobody can tell if their problem is listed. Status stays green during an obvious outage because updating it needed a manager's approval, which destroys trust faster than the outage. Fully automatic status changes from monitoring flap and post false outages, so most teams automate detection and alerts but keep a human posting the public status, with a short time target (for example 10 to 15 minutes after an incident is declared). Status levels need definitions so the choice between "degraded performance" and "partial outage" is not argued during the incident. Updates follow a cadence even with no news ("still investigating, next update in 30 minutes"). The page itself must be hosted separately from the product so it survives the outage it reports, and subscribers (email, RSS, webhook) reduce support load.
</context>
<task>
<product>
{{product}}
</product>
{{#setup}}
<setup>
{{setup}}
</setup>
{{/setup}}
1. Components: a list grouped as customers see the product (features, platforms, API, regions), 5 to 15 items, each with a one-line description in customer words and the internal services or checks behind it (for the team's reference, not shown publicly).
2. Status definitions: operational, degraded performance, partial outage, major outage, maintenance, each with a concrete criterion for this product.
3. Ownership and rules: who may post (the incident commander or a communications role), the time target for the first post after declaring an incident, update cadence per severity, no approval chain for the first post, and when to post for partial or single-region issues.
4. Templates: investigating, identified, monitoring, resolved, and scheduled maintenance, with placeholders, written in plain customer language without internal names or blame.
5. Automation: which monitoring signals should alert the team to update the page, what may be automated (scheduled maintenance, uptime display), subscriber notifications, and keeping the page hosted independently of production.
6. Launch checklist: steps to go live, including a test incident on a private page, linking the page from the help centre and app, and support team briefing.
</task>
<constraints>
- Use customer language in all public text; no internal service names, hostnames or vendor names unless the user wants them.
- Do not promise uptime numbers or SLA credits on the page; that belongs to the contract.
- Do not recommend a specific vendor as required; if the user has not chosen a tool, list the capabilities to look for.
- Keep templates honest: no "a small number of users" unless that is known.
{{> output/uncertainty}}
</constraints>
<output_format>
## Components
Table: Public component | Description | Internal mapping.
## Status definitions
Status | Criterion for this product.
## Ownership and rules
Bullets.
## Templates
One block per template.
## Automation
Bullets.
## Launch checklist
Checklist.
</output_format>可替换内容
- {{product}}
- {{#setup}}
- {{setup}}
- {{/setup}}
- {{> output/uncertainty}}
分类与标签
编程开发、状态页建设、故障通报、软件运维、事件响应