Back to Details

installer-tester

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

InstallerTester

InstallerTester

No tool gets installed without passing this check.
任何工具都必须通过此校验才能安装。

Requirements

要求

SkillRoleRequired?
installer-auditor
Post-install validationRecommended
skill-vetter
Security vetting for external skillsRequired if installing a skill
技能角色是否必需？
installer-auditor
安装后验证推荐
skill-vetter
外部技能的安全审核若安装技能则为必需

Triggers

触发条件

  • "install X", "add X", "set up X", "should we use X"
  • GitHub/npm/brew link shared
  • "is X worth installing", "compare X to Y"
  • /installer_tester
    , 
    /software-check
  • 「安装X」「添加X」「设置X」「我们是否应该使用X」
  • 分享GitHub/npm/brew链接
  • 「X是否值得安装」「对比X与Y」
  • /installer_tester
    , 
    /software-check

Phase 0: Docs & Security

阶段0：文档与安全

  1. Run 
    openclaw docs "<tool-name>"
    (fallback: https://docs.openclaw.ai). Informs Check 3.
  2. If target is a skill → activate 
    skill-vetter
    , get ✅ SAFE or ⚠️ CAUTION (with approval) before proceeding. Fallback: manually review SKILL.md for red flags.
  1. 运行
    openclaw docs "<tool-name>"
    （备选方案：https://docs.openclaw.ai）。结果将用于校验3。
  2. 若目标为技能 → 启动
    skill-vetter
    ，获取✅ SAFE（安全）或⚠️ CAUTION（需审批）后再继续。备选方案：手动检查SKILL.md中的风险信号。

Phase 1: Pre-Install Check

阶段1：安装前校验

Run in order. Stop and report if any check is a clear no.
按顺序执行。若任意校验明确不通过，则停止并上报。

Check 1 — What does it do?

校验1 — 工具功能？

State what it does in one sentence. Not what the user thinks — what it actually does. Can't determine? Stop.
用一句话说明该工具的实际功能，而非用户认为的功能。若无法确定，则停止。

Check 2 — What would it enhance or fix?

校验2 — 能改善或解决什么问题？

Search error logs, memory, session history, TOOLS.md, skills for:
  • Friction: errors, workarounds, slow steps, token waste
  • Enhancement: faster, cheaper, more reliable, new capability
Neither found → "Shiny toy — skip."
搜索错误日志、内存记录、会话历史、TOOLS.md、技能库，查找：
  • 痛点：错误、临时解决方案、缓慢步骤、令牌浪费
  • 优化点：更快、更廉价、更可靠、新功能
若均未找到 → 判定为「华而不实的工具——跳过」。

Check 3 — Already covered?

校验3 — 是否已覆盖需求？

Check docs results from Phase 0, then:
bash
mcporter list                                              # optional — requires mcporter
which <toolname> 2>/dev/null || echo "not installed"
openclaw skills list 2>&1 | grep -i "<toolname>"
grep -i "<toolname>" ~/.openclaw/workspace/TOOLS.md
Verdict: Not covered → Check 4 | Partially covered → Check 4 | Fully covered → stop, report what covers it.
先查看阶段0的文档结果，再执行以下命令：
bash
mcporter list                                              # 可选——需依赖mcporter
which <toolname> 2>/dev/null || echo "not installed"
openclaw skills list 2>&1 | grep -i "<toolname>"
grep -i "<toolname>" ~/.openclaw/workspace/TOOLS.md
判定结果：未覆盖 → 进入校验4 | 部分覆盖 → 进入校验4 | 完全覆盖 → 停止，上报现有覆盖工具。

Check 4 — Better than what we have?

校验4 — 比现有工具更优？

If overlap: quantify improvement (tokens, speed, reliability, capability).
Health check: last commit, stars, issues, bus factor, dependency count, red flags (archived, abandoned, no docs).
若存在功能重叠：量化改进点（令牌消耗、速度、可靠性、功能）。 健康检查：最后一次提交、星标数、问题数、核心维护者数量、依赖项数量、风险信号（归档、废弃、无文档）。

Decision Output

决策输出

One summary line: 
Coverage: [not covered/partial/fully covered] | Quality: active/stale/dead | Verdict: Install / Skip / Flag
Wait for approval.
总结格式： 
覆盖情况：[未覆盖/部分覆盖/完全覆盖] | 质量状态：活跃/停滞/废弃 | 结论：安装/跳过/标记
等待审批。

Phase 2: Install

阶段2：安装

Follow the tool's install instructions (Homebrew, npm, pip, 
claude mcp add
).
  • Dependency missing or install fails → stop, report. Don't improvise.
  • API keys → platform secure credential system. Never inline.
  • Record install method (needed for rollback).
遵循工具的安装说明（Homebrew、npm、pip、
claude mcp add
）。
  • 若依赖缺失或安装失败 → 停止并上报，请勿自行尝试解决。
  • API密钥 → 使用平台安全凭证系统，绝不硬编码。
  • 记录安装方式（用于后续回滚）。

Phase 3: Validation

阶段3：验证

If InstallerAuditor is a registered agent:
sessions_spawn(agentId: "installer_auditor", task: "Validate: [WHAT]. Method: [METHOD]. Expected: [DESCRIPTION].", label: "installer-auditor")
Otherwise read the 
installer-auditor
skill and follow its tiers directly.
Validation fails → uninstall, report.
若InstallerAuditor已注册为Agent：
sessions_spawn(agentId: "installer_auditor", task: "Validate: [WHAT]. Method: [METHOD]. Expected: [DESCRIPTION].", label: "installer-auditor")
否则，请查阅
installer-auditor
技能文档并直接遵循其验证层级要求。
若验证失败 → 卸载并上报。

Phase 4: Document

阶段4：文档记录

Log in TOOLS.md: name, purpose, invocation, priority order. Check OpenClaw docs first for native coverage.
在TOOLS.md中记录：名称、用途、调用方式、优先级顺序。先检查OpenClaw文档是否已有原生覆盖。

Completion Checklist

完成检查清单

  • Phase 0 — docs + vetter (if skill)
  • Check 1 — one-sentence description
  • Check 2 — friction/enhancement or "shiny toy"
  • Check 3 — coverage verdict with real checks
  • Check 4 — health assessed (if not fully covered)
  • Decision — verdict delivered, approval received
  • Phase 2 — installed (method recorded)
  • Phase 3 — validated
  • Phase 4 — documented
Unchecked without a reason = not done.
  • 阶段0 — 文档检查 + 技能审核（若安装技能）
  • 校验1 — 完成一句话功能描述
  • 校验2 — 找到痛点/优化点或判定为「华而不实的工具」
  • 校验3 — 完成覆盖情况判定及实际检查
  • 校验4 — 完成健康状态评估（若未完全覆盖需求）
  • 决策 — 给出结论并获得审批
  • 阶段2 — 完成安装（记录安装方式）
  • 阶段3 — 完成验证
  • 阶段4 — 完成文档记录
无正当理由未勾选 = 未完成。