跳到正文
cccskills
免费在 GitHub 公开

requirement-preflight

Review a product requirement, PRD, brief, or meeting note before design or development; expose missing decisions, contradictions, edge cases, dependencies, and validation gaps without inventing business rules.

中文简介正在准备中。当前显示原文说明。

实测评分 未实测

查看安装方法

包含的文件(2)

  • SKILL.md3.2 KB
  • agents/openai.yaml261 B

SKILL.md(原文)

安装前可以查看交给 AI 代理的指令内容。

Requirement Preflight

Turn an incoming requirement into an evidence-grounded readiness review that helps product, design, engineering, and QA decide what must be clarified before work proceeds.

Working rules

  • Treat supplied artifacts as the source of truth. Separate explicit facts, reasonable inferences, and unknowns.
  • Do not silently complete missing product decisions. Phrase them as assumptions or questions.
  • Distinguish a documentation gap from a flawed requirement; missing text alone does not prove the product decision is wrong.
  • Cite the relevant section, quote, screen, or source label when available. If provenance is unavailable, say so.
  • Prioritize issues by likely impact on scope, user harm, compliance, implementation, or rework—not by writing quality.
  • Match the language and depth to the user's audience. Preserve domain terminology from the source.
  • Do not modify source documents or external systems unless the user explicitly asks.

Review workflow

  1. Establish the review boundary: the supplied artifacts, intended stage, target users, and known constraints. If context is sparse, still review what exists and label the gaps.
  2. Restate the requirement briefly: user problem, business outcome, target users, core scenario, scope, and proposed success signal.
  3. Inspect the requirement across the dimensions that materially apply:
    • problem evidence and expected outcome;
    • users, roles, permissions, entry conditions, and exclusions;
    • happy path, alternate paths, recovery, cancellation, and reversibility;
    • loading, empty, error, partial-success, duplicate, timeout, and extreme-data states;
    • business rules, lifecycle states, calculations, limits, and conflicts;
    • data sources, integrations, dependencies, migration, and operational ownership;
    • privacy, security, accessibility, localization, and regulatory constraints;
    • analytics, release strategy, acceptance criteria, and validation plan.
  4. Classify findings:
    • Blocker: a decision is required before responsible design or implementation can continue.
    • High rework risk: work can start, but the uncertainty is likely to change scope or core flow.
    • Improvement: clarification would improve quality but need not block progress.
  5. Convert findings into answerable questions. Name the likely owner—product, design, engineering, data, legal, operations, or business—when it is evident.
  6. End with a readiness recommendation: ready, conditionally ready, or not ready, followed by the smallest next actions.

Default deliverable

Use a compact structure unless the user requests another format:

  1. Requirement snapshot
  2. Readiness recommendation and rationale
  3. Findings by severity, each with evidence, impact, and owner
  4. Missing flows and states
  5. Assumptions requiring confirmation
  6. Questions for the review meeting
  7. Suggested next actions

When comparing versions, add a change-impact section covering resolved questions, new risks, and affected designs, analytics, or acceptance criteria.

评价

还没有评价。欢迎分享你的使用感受。

评论

无需登录即可发表。欢迎分享使用感受、提问,或说明在哪些环境下无法运行。

还没有评论。来写第一条吧。

提交后会立即公开。请勿填写邮箱、电话等个人信息。URL 最多 2 个。

0 / 1000 字

此表单受 reCAPTCHA 保护,适用 Google 的隐私权政策和服务条款。

同一仓库的技能

Review product designs before engineering handoff or release for flow and state coverage, business-rule alignment, implementation clarity, accessibility, responsiveness, analytics, and testability.

中文简介正在准备中。当前显示原文说明。

MUZI-LYY/design-team-ai-skills92026年8月26日 更新

Synthesize user interviews, observations, surveys, support tickets, reviews, or feedback into traceable themes, insights, tensions, opportunities, and research gaps; use when evidence must inform product or design decisions.

中文简介正在准备中。当前显示原文说明。

MUZI-LYY/design-team-ai-skills92026年8月26日 更新

查看 MUZI-LYY 的全部技能

报告此技能的问题