Back to Details

review

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Review Orchestrator

审查编排器

When to use

使用场景

  • User wants a comprehensive review but doesn't know where to start
  • Pre-launch readiness assessment
  • Post-incident review planning
  • New team member wants to understand review coverage
  • 用户想要全面审查但不知从何入手
  • 上线前就绪评估
  • 事后审查规划
  • 新团队成员想要了解审查覆盖范围

How it works

工作原理

Three modes:
  • Interactive triage (default) — asks context questions, recommends a review plan
  • Full suite (
    /review --all
    ) — runs all applicable categories
  • Targeted — user invokes a specific review directly
三种模式：
  • 交互式分类（默认）——询问上下文问题，推荐审查计划
  • 全套件模式 (
    /review --all
    ) ——运行所有适用类别的审查
  • 定向模式——用户直接调用特定审查

Steps

步骤

1. Gather context

1. 收集上下文

  • Read the project: tech stack, framework, package.json, routes, components
  • Check git diff for recent changes
  • Look for existing review reports in 
    shiplight/reports/
  • Check for compliance markers (HIPAA mentions, PCI references, GDPR cookies)
  • 读取项目信息：技术栈、框架、package.json、路由、组件
  • 检查git diff查看近期变更
  • 查找
    shiplight/reports/
    中的现有审查报告
  • 检查合规标记（HIPAA提及、PCI参考、GDPR cookie）

2. Ask targeted questions (max 4)

2. 提出针对性问题（最多4个）

Ask one at a time, with auto-detected defaults:
  1. What type of application? (SaaS, healthcare, fintech, e-commerce, internal tool, marketing site, API-only)
  2. What triggered this review? (pre-launch, new feature, dependency update, security incident, audit prep, routine)
  3. Any compliance requirements? (none, HIPAA, SOC2, PCI-DSS, GDPR, multiple) — auto-detect from codebase
  4. Specific concerns? (open-ended, optional)
逐个提问，并提供自动检测的默认选项：
  1. 应用程序类型？（SaaS、医疗健康、金融科技、电子商务、内部工具、营销站点、仅API）
  2. 触发本次审查的原因？（上线前、新功能、依赖更新、安全事件、审计准备、例行检查）
  3. 是否有合规要求？（无、HIPAA、SOC2、PCI-DSS、GDPR、多项）——从代码库自动检测
  4. 特定关注点？（开放式问题，可选）

3. Generate review plan

3. 生成审查计划

Based on answers, categorize all 8 review types as:
  • CRITICAL — must run, high risk of issues
  • RECOMMENDED — should run, meaningful value
  • OPTIONAL — nice to have
Present the plan with rationale for each recommendation. Include estimated depth (quick/standard/thorough) for each.
SEO vs GEO prioritization by product type:
Product typeSEOGEO
Developer tools, API products, SaaSRECOMMENDEDCRITICAL
E-commerce, local business, marketplaceCRITICALOPTIONAL
Content/media, documentation, blogCRITICALCRITICAL
Internal tools——
Provide a decision matrix table:
ReviewPriorityRationaleDepth
/security-reviewCRITICALNew auth feature + SaaS appthorough
/privacy-reviewCRITICALHandles user PII, GDPR appliesstandard
etc.
根据回答，将所有8种审查类型分类为：
  • CRITICAL（关键）——必须运行，存在高风险问题
  • RECOMMENDED（推荐）——应该运行，具有重要价值
  • OPTIONAL（可选）——锦上添花
呈现计划时需说明每个推荐的理由。包含每个审查的预计深度（快速/标准/全面）。
按产品类型区分SEO与GEO的优先级：
产品类型SEOGEO
开发者工具、API产品、SaaSRECOMMENDEDCRITICAL
电子商务、本地商家、市场平台CRITICALOPTIONAL
内容/媒体、文档、博客CRITICALCRITICAL
内部工具——
提供决策矩阵表格：
审查优先级理由深度
/security-reviewCRITICAL新增认证功能 + SaaS应用thorough
/privacy-reviewCRITICAL处理用户PII，适用GDPRstandard
etc.

4. Execute

4. 执行

Ask: "Run all CRITICAL reviews now? [Y/n] Or pick specific ones."
Run selected reviews sequentially. After each, show a brief summary before proceeding to the next.
询问："立即运行所有CRITICAL审查？[Y/n] 或选择特定审查。"
按顺序运行选定的审查。每次审查完成后，显示简要摘要再进行下一个。

5. Unified report

5. 统一报告

After all reviews complete, generate a unified report:
  • Overall readiness score (0-10)
  • Per-category scores
  • Top 5 findings across all categories (by severity)
  • Regression test summary (total YAML tests generated)
  • Report saved to 
    shiplight/reports/review-{date}.md
所有审查完成后，生成统一报告：
  • 整体就绪评分（0-10分）
  • 各分类评分
  • 所有类别中最严重的5项发现（按严重性排序）
  • 回归测试摘要（生成的YAML测试总数）
  • 报告保存至
    shiplight/reports/review-{date}.md

Available Reviews

可用审查

SkillCategoryWhat it checks
/design-review
Visual/UIResponsive, a11y, design consistency, i18n readiness
/security-review
SecurityOWASP Top 10, auth, headers, supply chain, pen testing
/privacy-review
PrivacyPII handling, tracking, data flow, consent
/compliance-review
ComplianceHIPAA, SOC2, PCI-DSS, GDPR checklists
/resilience-review
ReliabilityError handling, degradation, API contracts
/performance-review
PerformanceCore Web Vitals, bundle size, runtime perf
/seo-review
DiscoverabilityMeta tags, structured data, crawlability
/geo-review
AI DiscoverabilityLLM citation readiness, entity authority, structured claims
Skill分类检查内容
/design-review
视觉/UI响应式设计、无障碍（a11y）、设计一致性、国际化（i18n）就绪情况
/security-review
安全OWASP Top 10、认证、请求头、供应链、渗透测试
/privacy-review
隐私PII处理、追踪、数据流、用户同意
/compliance-review
合规HIPAA、SOC2、PCI-DSS、GDPR检查清单
/resilience-review
可靠性错误处理、降级机制、API契约
/performance-review
性能Core Web Vitals、包大小、运行时性能
/seo-review
可发现性元标签、结构化数据、可抓取性
/geo-review
AI可发现性LLM引用就绪情况、实体权威性、结构化声明

Report Format

报告格式

All review skills produce reports in a consistent format saved to 
shiplight/reports/{review-name}-{date}.md
. The orchestrator merges these into a unified report.
所有审查Skill生成的报告均采用一致格式，保存至
shiplight/reports/{review-name}-{date}.md
。编排器会将这些报告合并为一份统一报告。

Tips

提示

  • Run 
    /review
    before every major launch
  • Individual reviews can be invoked directly when you know what you need
  • Review reports accumulate over time — the orchestrator can show trends
  • YAML regression tests from reviews accumulate in 
    shiplight/tests/
  • 每次重大上线前运行
    /review
  • 当您明确需求时，可以直接调用单个审查
  • 审查报告随时间累积——编排器可显示趋势
  • 审查生成的YAML回归测试会累积在
    shiplight/tests/
    中