Translate subjective, vague user requirements into structured, precise AI-executable specifications.
Trigger Conditions
This Skill is triggered by any of the following situations:
- The user's input requirement is ambiguous, vaguely described, or purely subjective (such as "good-looking", "easy to use", "a bit faster")
- The user explicitly requests "requirement translation", "translate requirements", "clarify requirements", "refine requirements"
- The user provides a requirement but lacks key details (tech stack, scope, acceptance criteria, boundary conditions)
- The user gives instructions like "make a good-looking XX", "help me optimize YY", "implement a ZZ feature", etc., where the goal is unclear but inferable
Workflow
Phase 1 — Requirement Analysis
Extract from the original requirement:
- Core Goal: What does the user really want to achieve?
- Ambiguity Points: Which statements are ambiguous or open to multiple interpretations?
- Missing Information: Which key parameters are missing (tech stack, scope, constraints, etc.)?
- Implicit Assumptions: What might the user have assumed by default but not stated explicitly?
Phase 2 — Translation Output
Translate the requirement into the following structured format. If a section has no content, note "None" rather than omitting it.
## 需求规格书
### 1. 目标概述
[一句话描述核心目标，点明要解决的问题和预期结果]
### 2. 功能需求
- F1: [具体可执行的功能描述]
- F2: [具体可执行的功能描述]
### 3. 技术约束
- T1: [技术栈 / 平台 / 兼容性要求]
- T2: [性能 / 安全 / 规模约束]
### 4. 设计约束
- D1: [视觉风格 / 交互模式 / 品牌约束]
- D2: [布局 / 响应式 / 无障碍要求]
### 5. 验收标准
- AC1: [可客观验证的完成标准，含具体数值或状态]
- AC2: [可客观验证的完成标准]
### 6. 边界条件与异常处理
- E1: [异常场景及预期处理方式]
- E2: [极端输入 / 空状态 / 错误状态处理]
### 7. 优先级
- P0 必须: [核心不可妥协的需求]
- P1 应该: [重要但可后续迭代]
- P2 可选: [锦上添花，可省略]
Phase 3 — Execution Recommendations
Append after the translation result:
- Recommended Approach: Recommended technical route and rationale (1-2 sentences)
- Risk Alerts: Potential pitfalls, technical difficulties, or uncertainties
- Implementation Order: Recommended step-by-step execution path
Working Principles
- Ask When Not Omniscient: When key information is missing and cannot be reasonably inferred, list the questions to be confirmed; do not guess
- Reasonable Inference: For non-key information that can be inferred based on context and industry conventions, adopt it directly and annotate it
[推断] - Executability First: Each requirement item must be executable and verifiable; avoid unverifiable expressions such as "do it well" or "do it right"
- Preserve Intent: Do not forcibly change the user's original intent; only make it precise and structured
- Inference Annotation: Mark all inferred content with
[推断]so the user can review and correct it - Chinese Output: Always output the translation result in Chinese