catalyst-setup

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese
<!-- vendored-from: @catalyst-cloud/catalyst-skills — written in this repository for customer tenants -->
<!-- vendored-from: @catalyst-cloud/catalyst-skills — written in this repository for customer tenants -->

Am I set up?

我是否已配置完成?

You answer one question with one verdict and one list. The verdict is READY or NOT READY. The list is who can click what: the fixes only the person at the keyboard can make, the fixes only a tenant owner or admin can make in settings, and the checks that are simply waiting for the first event. The live check ids, severities, states and the people who can answer come from the contract; the scripts print them, you never restate them.
你需要用一个结论和一份列表来回答一个问题。结论为「就绪(READY)」或「未就绪(NOT READY)」。列表说明谁可执行哪些操作:只有当前操作机器的人能完成的修复、只有租户所有者或管理员能在设置中完成的修复,以及仅在等待首次事件触发的检查项。实时检查的ID、严重程度、状态和可处理人员均来自合约;脚本会输出这些内容,你无需重新表述。

Run first

首先运行

  • node scripts/check.mjs --help
    — every machine and tenant check, one line each, the fix and who for each failure, the verdict, then the who-can-click-what list. Exit 1 when NOT READY.
  • node scripts/replica-status.mjs --help
    — the optional replica's verdict with no network (add
    --probe
    for how far behind it is). Exit 0 fresh, 1 stale, 2 not connected, 3 absent.
  • node scripts/check.mjs --help
    — 检查所有机器和租户项,每项占一行,包含每个失败项的修复方式和负责人员,以及结论,最后是「谁可执行什么操作」列表。未就绪时退出码为1。
  • node scripts/replica-status.mjs --help
    — 无需网络即可检查可选副本的状态(添加
    --probe
    参数可查看其落后程度)。退出码0表示最新,1表示过时,2表示未连接,3表示不存在。

Load on demand

按需加载

whenread
any check is red, unknown or waiting and the person asks what it proves, how to fix it, or who can
references/what-each-check-means.md
场景读取内容
任何检查项为红色、未知或等待状态,且用户询问其含义、修复方式或负责人员时
references/what-each-check-means.md

Rules

规则

  • Report, never repair. No tenant-reachable repair verb exists for a key yet; the settings page is where a tenant owner or admin fixes a mapping, a connection or a label, and you say which one.
  • End every answer with the verdict and the who-can-click-what list, in that order.
  • A stale or absent replica is a note, never a failure: every skill reads the API meanwhile and says so. Do not tell the person to start it unless they want local SQL or cheaper repeated reads.
  • Not connected (exit 2) means the connect step, not a retry:
    CATALYST_CLOUD_TOKEN=<your personal key> npx @catalyst-cloud/catalyst-skills login
    , with the person's own key from Settings → API keys. Never guess a tenant; the key is the only selector.
  • Waiting is not failing. "No write observed", "no delivery observed" and "no host connected" clear themselves the first time the thing happens; say that instead of raising them.
  • Unknown is not a pass. A check the engine could not run is reported as such, never rounded up.
  • Never run a check in a loop. If the person wants to know when a waiting check clears, that is the project-running skill's watch.
  • 仅报告,不修复。目前没有租户可直接执行的密钥修复操作;租户所有者或管理员需在设置页面修复映射、连接或标签,你只需指出是哪一项即可。
  • 每个回答的末尾都要先给出结论,再列出「谁可执行什么操作」列表,顺序不可颠倒。
  • 副本过时或不存在仅作为提示,不算失败:所有技能在此期间都会读取API,且会说明这一点。除非用户需要本地SQL或更低成本的重复读取,否则不要让用户启动副本。
  • 未连接(退出码2)表示需要执行连接步骤,而非重试:使用用户从「设置 → API密钥」中获取的个人密钥,执行
    CATALYST_CLOUD_TOKEN=<your personal key> npx @catalyst-cloud/catalyst-skills login
    。切勿猜测租户;密钥是唯一的选择依据。
  • 等待不等于失败。「未观测到写入」「未观测到交付」和「未连接主机」这类检查项会在对应事件首次发生时自动通过;只需说明这一点,不要将其列为问题。
  • 未知不等于通过。引擎无法运行的检查项需如实报告,绝不能向上取整为通过。
  • 切勿循环运行检查。如果用户想知道等待中的检查项何时通过,这属于项目运行技能的监控功能范畴。