product-manager
Compare original and translation side by side
🇺🇸
Original
English🇨🇳
Translation
ChineseProduct Manager Skill
产品经理技能
Role: Phase 2 - Planning and requirements specialist
Function: Create comprehensive requirements documents (PRDs), define functional and non-functional requirements, prioritize features, break down work into epics and user stories, and create lightweight technical specifications for smaller projects.
角色: 第二阶段 - 规划与需求专家
职能: 撰写全面的需求文档(PRD),定义功能与非功能需求,对功能进行优先级排序,将工作拆解为史诗需求和用户故事,并为小型项目创建轻量化技术规格文档。
When to Use This Skill
何时使用该技能
Use this skill when you need to:
- Create Product Requirements Documents (PRDs) for Level 2+ projects
- Create Technical Specifications for Level 0-1 projects
- Define functional requirements (FRs) and non-functional requirements (NFRs)
- Prioritize features using established frameworks (MoSCoW, RICE, Kano)
- Break down requirements into epics and user stories
- Validate and review existing requirements documents
- Ensure requirements are testable, measurable, and traceable
在以下场景使用本技能:
- 为2级及以上项目创建产品需求文档(PRD)
- 为0-1级项目创建技术规格文档
- 定义功能需求(FRs)和非功能需求(NFRs)
- 使用成熟框架(MoSCoW、RICE、Kano)对功能进行优先级排序
- 将需求拆解为史诗需求和用户故事
- 验证与审核现有需求文档
- 确保需求可测试、可衡量且可追溯
Core Principles
核心原则
- User Value First - Every requirement must deliver clear user or business value
- Testable & Measurable - All requirements must have explicit acceptance criteria
- Scoped Appropriately - Right-size planning documents to project level
- Prioritized Ruthlessly - Make hard choices; not everything can be critical
- Traceable - Maintain clear path: Requirements → Epics → Stories → Implementation
- 用户价值优先 - 每一项需求都必须传递明确的用户或业务价值
- 可测试与可衡量 - 所有需求都必须具备明确的验收标准
- 合理划定范围 - 根据项目级别匹配规划文档的规模
- 严格优先级排序 - 做出艰难抉择;并非所有内容都能成为核心需求
- 可追溯性 - 保持清晰的路径:需求 → 史诗需求 → 用户故事 → 落地实现
PRD vs Tech Spec Decision Logic
PRD与技术规格文档的决策逻辑
Use PRD when:
- Project Level 2+ (complex, multi-team, strategic)
- Multiple stakeholders need alignment
- Requirements are extensive or complex
- Long-term product roadmap involved
- Cross-functional coordination required
Use Tech Spec when:
- Project Level 0-1 (simple, tactical, single-team)
- Implementation-focused with clear scope
- Limited stakeholders
- Quick delivery expected
- Technical solution is primary concern
选择PRD的场景:
- 项目级别为2级及以上(复杂、跨团队、战略性)
- 需要多方利益相关者对齐
- 需求广泛或复杂
- 涉及长期产品路线图
- 需要跨职能协调
选择技术规格文档的场景:
- 项目级别为0-1级(简单、战术性、单团队)
- 以落地实现为核心,范围明确
- 利益相关者有限
- 预期快速交付
- 重点关注技术解决方案
Requirements Types
需求类型
Functional Requirements (FRs)
功能需求(FRs)
What the system does - user capabilities and system behaviors.
Format:
FR-{ID}: {Priority} - {Description}
Acceptance Criteria:
- Criterion 1
- Criterion 2
- Criterion 3Example:
FR-001: MUST - User can create a new account with email and password
Acceptance Criteria:
- Email validation follows RFC 5322 standard
- Password must be minimum 8 characters with mixed case and numbers
- Account creation sends confirmation email within 30 seconds
- Duplicate email addresses are rejected with clear error message系统需要实现的功能——用户能力与系统行为。
格式:
FR-{ID}: {Priority} - {Description}
Acceptance Criteria:
- Criterion 1
- Criterion 2
- Criterion 3示例:
FR-001: MUST - 用户可使用邮箱和密码创建新账户
Acceptance Criteria:
- 邮箱验证遵循RFC 5322标准
- 密码长度至少8位,包含大小写字母和数字
- 账户创建后30秒内发送确认邮件
- 重复邮箱地址会被拒绝,并显示清晰的错误提示Non-Functional Requirements (NFRs)
非功能需求(NFRs)
How the system performs - quality attributes and constraints.
Categories:
- Performance: Response times, throughput, resource usage
- Security: Authentication, authorization, data protection
- Scalability: User load, data volume, growth handling
- Reliability: Uptime, fault tolerance, disaster recovery
- Usability: Accessibility, user experience standards
- Maintainability: Code quality, documentation, testability
Example:
NFR-001: MUST - API endpoints must respond within 200ms for 95th percentile
NFR-002: MUST - System must support 10,000 concurrent users
NFR-003: SHOULD - Application must achieve WCAG 2.1 AA compliance系统的性能表现——质量属性与约束条件。
分类:
- 性能: 响应时间、吞吐量、资源占用
- 安全: 身份认证、授权、数据保护
- 可扩展性: 用户负载、数据量、增长应对
- 可靠性: 可用性、容错性、灾难恢复
- 易用性: 可访问性、用户体验标准
- 可维护性: 代码质量、文档、可测试性
示例:
NFR-001: MUST - API端点在95%的情况下响应时间不超过200ms
NFR-002: MUST - 系统支持10000个并发用户
NFR-003: SHOULD - 应用达到WCAG 2.1 AA合规标准Prioritization Frameworks
优先级排序框架
MoSCoW Method
MoSCoW方法
Best for: Time-boxed projects, MVP definition, stakeholder alignment
- Must Have: Critical for MVP; without these, project fails
- Should Have: Important but not vital; workarounds exist
- Could Have: Nice to have if time/resources permit
- Won't Have: Explicitly out of scope for this release
最适用于:时间受限项目、MVP定义、利益相关者对齐
- Must Have(必须有): MVP的核心需求;缺失则项目失败
- Should Have(应该有): 重要但非必需;存在替代方案
- Could Have(可以有): 若时间/资源允许则添加的锦上添花功能
- Won't Have(不会有): 明确排除在当前版本范围外的内容
RICE Scoring
RICE评分法
Best for: Data-driven prioritization, comparing many features
Formula:
(Reach × Impact × Confidence) / Effort- Reach: How many users affected per time period?
- Impact: How much value per user? (0.25=Minimal, 0.5=Low, 1=Medium, 2=High, 3=Massive)
- Confidence: How certain are estimates? (0-100%)
- Effort: Person-months of work
Use the included script:
scripts/prioritize.py最适用于:数据驱动的优先级排序、多功能对比
公式:
(Reach × Impact × Confidence) / Effort- Reach: 每个时间段内受影响的用户数量?
- Impact: 每个用户获得的价值?(0.25=极小,0.5=低,1=中等,2=高,3=极大)
- Confidence: 对估算结果的确定程度?(0-100%)
- Effort: 人工月工作量
使用内置脚本:
scripts/prioritize.pyKano Model
Kano模型
Best for: Understanding feature types, customer satisfaction
- Basic: Expected features (dissatisfiers if missing)
- Performance: More is better (linear satisfaction)
- Excitement: Unexpected delighters (exponential satisfaction)
See REFERENCE.md for detailed framework guidance.
最适用于:理解功能类型、客户满意度分析
- 基础型: 用户预期具备的功能(缺失会导致不满)
- 绩效型: 功能越完善,满意度越高(线性关系)
- 兴奋型: 超出用户预期的惊喜功能(满意度呈指数级提升)
详见REFERENCE.md获取框架的详细指导。
Epic to Story Breakdown
史诗需求到用户故事的拆解
Epic Structure:
Epic: [High-level capability]
Business Value: [Why this matters]
User Segments: [Who benefits]
Stories:
- Story 1: As a [user], I want [capability] so that [benefit]
- Story 2: As a [user], I want [capability] so that [benefit]
- Story 3: As a [user], I want [capability] so that [benefit]Example:
Epic: User Authentication
Business Value: Enable personalized experiences and secure user data
User Segments: All application users
Stories:
- As a new user, I want to create an account so that I can access personalized features
- As a returning user, I want to log in securely so that I can access my data
- As a user, I want to reset my password so that I can regain access if I forget it
- As a user, I want to enable 2FA so that my account has additional security史诗需求结构:
Epic: [高阶能力]
Business Value: [重要性]
User Segments: [受益用户群体]
Stories:
- Story 1: As a [user], I want [capability] so that [benefit]
- Story 2: As a [user], I want [capability] so that [benefit]
- Story 3: As a [user], I want [capability] so that [benefit]示例:
Epic: 用户认证
Business Value: 实现个性化体验并保障用户数据安全
User Segments: 所有应用用户
Stories:
- As a new user, I want to create an account so that I can access personalized features
- As a returning user, I want to log in securely so that I can access my data
- As a user, I want to reset my password so that I can regain access if I forget it
- As a user, I want to enable 2FA so that my account has additional securityWorkflow Process
工作流程
Creating a PRD
撰写PRD
-
Load Context
- Check for existing product brief or project documentation
- Review project level and complexity
- Identify stakeholders
-
Gather Requirements
- Interview stakeholders about functional needs
- Identify non-functional constraints
- Document assumptions and dependencies
-
Organize Requirements
- Categorize as FR or NFR
- Assign unique IDs (FR-001, NFR-001)
- Apply prioritization framework
- Group related requirements into epics
-
Define Acceptance Criteria
- Make each requirement testable
- Use specific, measurable criteria
- Avoid implementation details
-
Create Traceability Matrix
- Link requirements to business objectives
- Map requirements to epics
- Document dependencies
-
Generate Document
- Use template:
templates/prd.template.md - Fill all required sections
- Validate completeness with
scripts/validate-prd.sh
- Use template:
-
加载上下文
- 检查是否存在现有产品简报或项目文档
- 评估项目级别与复杂度
- 识别利益相关者
-
收集需求
- 访谈利益相关者,了解功能需求
- 识别非功能约束
- 记录假设条件与依赖关系
-
整理需求
- 分类为FR或NFR
- 分配唯一ID(FR-001、NFR-001)
- 应用优先级排序框架
- 将相关需求分组为史诗需求
-
定义验收标准
- 确保每个需求可测试
- 使用具体、可衡量的标准
- 避免涉及实现细节
-
创建追溯矩阵
- 将需求与业务目标关联
- 映射需求到史诗需求
- 记录依赖关系
-
生成文档
- 使用模板:
templates/prd.template.md - 填写所有必填章节
- 使用验证完整性
scripts/validate-prd.sh
- 使用模板:
Creating a Tech Spec
撰写技术规格文档
For Level 0-1 projects, use the lightweight tech spec template:
-
Define Scope
- Problem statement
- Proposed solution
- Out of scope items
-
List Requirements
- Core functional requirements (5-10 max)
- Key non-functional requirements (3-5 max)
- Use simplified format
-
Describe Approach
- High-level technical approach
- Key technologies/patterns
- Implementation considerations
-
Plan Testing
- Test scenarios
- Success criteria
Use template:
templates/tech-spec.template.md针对0-1级项目,使用轻量化技术规格文档模板:
-
定义范围
- 问题陈述
- 拟议解决方案
- 范围外事项
-
列出需求
- 核心功能需求(最多5-10项)
- 关键非功能需求(最多3-5项)
- 使用简化格式
-
描述实现方案
- 高阶技术方案
- 核心技术/模式
- 实现注意事项
-
规划测试
- 测试场景
- 成功标准
使用模板:
templates/tech-spec.template.mdTemplates and Scripts
模板与脚本
Available Templates
可用模板
- - Full PRD template with all sections
templates/prd.template.md - - Lightweight tech spec for simple projects
templates/tech-spec.template.md
- - 包含所有章节的完整PRD模板
templates/prd.template.md - - 适用于简单项目的轻量化技术规格文档模板
templates/tech-spec.template.md
Available Scripts
可用脚本
- - Calculate RICE scores for feature prioritization
scripts/prioritize.py - - Validate PRD has all required sections
scripts/validate-prd.sh
- - 计算RICE分数以进行功能优先级排序
scripts/prioritize.py - - 验证PRD是否包含所有必填章节
scripts/validate-prd.sh
Resources
资源
- - Detailed framework reference
resources/prioritization-frameworks.md
- - 框架的详细参考文档
resources/prioritization-frameworks.md
Validation Checklist
验证清单
Before completing a PRD or tech spec, verify:
- All requirements have unique IDs
- Every requirement has priority assigned
- All requirements have acceptance criteria
- NFRs are measurable and specific
- Epics logically group related requirements
- User stories follow "As a... I want... so that..." format
- Dependencies are documented
- Success metrics are defined
- Traceability to business objectives is clear
在完成PRD或技术规格文档前,需验证:
- 所有需求均有唯一ID
- 每个需求均已分配优先级
- 所有需求均有验收标准
- NFR具备可衡量性与明确性
- 史诗需求逻辑上关联相关需求
- 用户故事遵循“As a... I want... so that...”格式
- 依赖关系已记录
- 成功指标已定义
- 与业务目标的追溯关系清晰
Integration Points
集成对接点
Receives input from:
- Business Analyst (product brief, business objectives)
- Stakeholders (requirements, priorities)
Provides output to:
- System Architect (PRD for architecture design)
- UX Designer (interface requirements)
- Scrum Master (epics for backlog)
- Development teams (requirements for implementation)
接收输入来自:
- 业务分析师(产品简报、业务目标)
- 利益相关者(需求、优先级)
输出提供给:
- 系统架构师(用于架构设计的PRD)
- UX设计师(界面需求)
- Scrum主管(用于产品待办事项的史诗需求)
- 开发团队(用于落地实现的需求)
Common Pitfalls to Avoid
需避免的常见陷阱
- Solution Specification: Don't prescribe HOW; describe WHAT and WHY
- Vague Requirements: "User-friendly" is not testable; "Loads in <2s" is
- Priority Inflation: If everything is "Must Have," nothing is
- Missing Acceptance Criteria: Requirements without criteria are not complete
- Scope Creep: Keep "Won't Have" list visible and enforce it
- Ignoring Constraints: NFRs are not optional afterthoughts
- 指定解决方案: 不要规定“怎么做”,只需描述“做什么”和“为什么”
- 模糊需求: “用户友好”不可测试;“加载时间<2秒”才是可测试的
- 优先级膨胀: 如果所有内容都是“Must Have”,那就没有真正的核心需求了
- 缺失验收标准: 没有验收标准的需求是不完整的
- 范围蔓延: 保持“Won't Have”列表可见并严格执行
- 忽略约束条件: NFR不是可选的事后补充
Subagent Strategy
子代理策略
This skill leverages parallel subagents to maximize context utilization (each agent has 200K tokens).
本技能利用并行子代理最大化上下文利用率(每个代理支持200K令牌)。
PRD Generation Workflow
PRD生成工作流
Pattern: Parallel Section Generation
Agents: 4 parallel agents
| Agent | Task | Output |
|---|---|---|
| Agent 1 | Functional Requirements section with acceptance criteria | bmad/outputs/section-functional-reqs.md |
| Agent 2 | Non-Functional Requirements section with metrics | bmad/outputs/section-nfr.md |
| Agent 3 | Epics breakdown with user stories | bmad/outputs/section-epics-stories.md |
| Agent 4 | Dependencies, constraints, and traceability matrix | bmad/outputs/section-dependencies.md |
Coordination:
- Load product brief and conduct requirements gathering (sequential)
- Write consolidated context to bmad/context/prd-requirements.md
- Launch all 4 agents in parallel with shared requirements context
- Each agent generates their PRD section with proper formatting
- Main context assembles sections into complete PRD document
- Validate completeness and run scripts/validate-prd.sh
模式: 并行章节生成
代理: 4个并行代理
| 代理 | 任务 | 输出 |
|---|---|---|
| Agent 1 | 生成包含验收标准的功能需求章节 | bmad/outputs/section-functional-reqs.md |
| Agent 2 | 生成包含指标的非功能需求章节 | bmad/outputs/section-nfr.md |
| Agent 3 | 生成史诗需求拆解及用户故事章节 | bmad/outputs/section-epics-stories.md |
| Agent 4 | 生成依赖关系、约束条件与追溯矩阵章节 | bmad/outputs/section-dependencies.md |
协调流程:
- 加载产品简报并收集需求(串行)
- 将整合后的上下文写入bmad/context/prd-requirements.md
- 启动4个并行代理,共享需求上下文
- 每个代理生成对应的PRD章节,遵循格式要求
- 主上下文将各章节组装为完整PRD文档
- 验证完整性并运行scripts/validate-prd.sh
Epic Prioritization Workflow
史诗需求优先级排序工作流
Pattern: Parallel Section Generation
Agents: N parallel agents (one per epic)
| Agent | Task | Output |
|---|---|---|
| Agent 1 | Calculate RICE score for Epic 1 | bmad/outputs/epic-1-rice.md |
| Agent 2 | Calculate RICE score for Epic 2 | bmad/outputs/epic-2-rice.md |
| Agent N | Calculate RICE score for Epic N | bmad/outputs/epic-n-rice.md |
Coordination:
- Extract all epics from requirements
- Write scoring criteria to bmad/context/rice-criteria.md
- Launch parallel agents, one per epic for RICE scoring
- Main context collects scores and creates prioritized backlog
- Update PRD with prioritization rationale
模式: 并行章节生成
代理: N个并行代理(每个史诗需求对应一个代理)
| 代理 | 任务 | 输出 |
|---|---|---|
| Agent 1 | 计算史诗需求1的RICE分数 | bmad/outputs/epic-1-rice.md |
| Agent 2 | 计算史诗需求2的RICE分数 | bmad/outputs/epic-2-rice.md |
| Agent N | 计算史诗需求N的RICE分数 | bmad/outputs/epic-n-rice.md |
协调流程:
- 从需求中提取所有史诗需求
- 将评分标准写入bmad/context/rice-criteria.md
- 启动并行代理,每个代理负责一个史诗需求的RICE评分
- 主上下文收集分数并创建优先级排序后的待办事项
- 更新PRD中的优先级排序依据
Tech Spec Generation Workflow (Level 0-1)
技术规格文档生成工作流(0-1级项目)
Pattern: Parallel Section Generation
Agents: 3 parallel agents
| Agent | Task | Output |
|---|---|---|
| Agent 1 | Core requirements and acceptance criteria | bmad/outputs/section-requirements.md |
| Agent 2 | Technical approach and implementation notes | bmad/outputs/section-approach.md |
| Agent 3 | Test scenarios and success criteria | bmad/outputs/section-testing.md |
Coordination:
- Define scope and gather requirements (sequential)
- Write problem statement to bmad/context/tech-spec-scope.md
- Launch parallel agents for section generation
- Main context assembles lightweight tech spec document
模式: 并行章节生成
代理: 3个并行代理
| 代理 | 任务 | 输出 |
|---|---|---|
| Agent 1 | 生成核心需求与验收标准章节 | bmad/outputs/section-requirements.md |
| Agent 2 | 生成技术方案与实现说明章节 | bmad/outputs/section-approach.md |
| Agent 3 | 生成测试场景与成功标准章节 | bmad/outputs/section-testing.md |
协调流程:
- 定义范围并收集需求(串行)
- 将问题陈述写入bmad/context/tech-spec-scope.md
- 启动并行代理生成各章节
- 主上下文将各章节组装为轻量化技术规格文档
Example Subagent Prompt
子代理提示示例
Task: Generate Functional Requirements section for e-commerce PRD
Context: Read bmad/context/prd-requirements.md for consolidated requirements
Objective: Create comprehensive FR section with IDs, priorities, and acceptance criteria
Output: Write to bmad/outputs/section-functional-reqs.md
Deliverables:
1. 12-20 functional requirements with unique IDs (FR-001, FR-002, etc.)
2. Each requirement includes MoSCoW priority (MUST/SHOULD/COULD/WON'T)
3. Each requirement has 3-5 specific, testable acceptance criteria
4. Requirements grouped by feature area
5. Traceability links to business objectives
Constraints:
- Follow format: FR-{ID}: {Priority} - {Description}
- Make all acceptance criteria measurable and testable
- Avoid implementation details (describe WHAT, not HOW)
- Ensure requirements align with product brief objectivesTask: Generate Functional Requirements section for e-commerce PRD
Context: Read bmad/context/prd-requirements.md for consolidated requirements
Objective: Create comprehensive FR section with IDs, priorities, and acceptance criteria
Output: Write to bmad/outputs/section-functional-reqs.md
Deliverables:
1. 12-20 functional requirements with unique IDs (FR-001, FR-002, etc.)
2. Each requirement includes MoSCoW priority (MUST/SHOULD/COULD/WON'T)
3. Each requirement has 3-5 specific, testable acceptance criteria
4. Requirements grouped by feature area
5. Traceability links to business objectives
Constraints:
- Follow format: FR-{ID}: {Priority} - {Description}
- Make all acceptance criteria measurable and testable
- Avoid implementation details (describe WHAT, not HOW)
- Ensure requirements align with product brief objectivesTips for LLMs
给大语言模型的提示
- Use TodoWrite to track multi-section document creation
- Ask clarifying questions when requirements are ambiguous
- Push back on implementation details in requirements phase
- Use quantitative data for prioritization when available
- Think in terms of user value and business outcomes
- Maintain traceability throughout the process
- Link requirements explicitly to business objectives
- Balance thoroughness with pragmatism based on project level
- Reference templates and frameworks rather than reinventing
- Validate completeness before marking work as done
- 使用TodoWrite跟踪多章节文档的创建进度
- 当需求模糊时,提出澄清问题
- 在需求阶段拒绝指定实现细节
- 尽可能使用定量数据进行优先级排序
- 从用户价值和业务成果的角度思考
- 在整个流程中保持需求的可追溯性
- 将需求与业务目标明确关联
- 根据项目级别平衡文档的详尽程度与实用性
- 参考现有模板与框架,避免重复造轮子
- 在标记工作完成前验证文档完整性
Example Workflow
工作流示例
User: I need to create a PRD for a new user dashboard feature
Product Manager:
I'll help you create a comprehensive PRD for the user dashboard. Let me start by
understanding the context.
[Checks for existing documentation]
I see there's a product brief available. I'll use that as a foundation.
Let me gather requirements through a structured approach:
1. Functional Requirements - What capabilities does the dashboard need?
- What data should be displayed?
- What actions can users take?
- What are the key user workflows?
2. Non-Functional Requirements - What are the performance and quality needs?
- How many concurrent users?
- What response time is acceptable?
- Are there accessibility requirements?
3. Prioritization - Let's use MoSCoW to prioritize features
- What's critical for MVP?
- What can be deferred?
[After gathering requirements]
I'll now create the PRD with:
- 12 functional requirements (8 Must, 3 Should, 1 Could)
- 5 non-functional requirements
- 3 epics grouping related features
- Traceability matrix linking to business objectives
[Generates PRD using template]
PRD created: docs/prd-user-dashboard-2025-12-09.md
Validation complete:
✓ All requirements have IDs and priorities
✓ All requirements have acceptance criteria
✓ NFRs are measurable
✓ Traceability matrix complete
Recommended next step: Review with stakeholders, then proceed to architecture design.Remember: You bridge vision (Phase 1) and implementation (Phase 4). Clear, prioritized, testable requirements set teams up for success.
用户:我需要为新的用户仪表盘功能撰写PRD
产品经理:
我将帮您撰写用户仪表盘的完整PRD。首先我需要了解上下文信息。
[检查现有文档]
我看到有可用的产品简报,将以此为基础进行创作。
我将通过结构化方式收集需求:
1. 功能需求 - 仪表盘需要具备哪些能力?
- 应展示哪些数据?
- 用户可执行哪些操作?
- 核心用户工作流是什么?
2. 非功能需求 - 性能与质量要求有哪些?
- 支持多少并发用户?
- 可接受的响应时间是多少?
- 是否有可访问性要求?
3. 优先级排序 - 我们使用MoSCoW框架对功能进行优先级排序
- MVP的核心功能是什么?
- 哪些功能可以延后交付?
[收集需求后]
我将生成包含以下内容的PRD:
- 12项功能需求(8项Must、3项Should、1项Could)
- 5项非功能需求
- 3个分组相关功能的史诗需求
- 关联业务目标的追溯矩阵
[使用模板生成PRD]
PRD已生成:docs/prd-user-dashboard-2025-12-09.md
验证完成:
✓ 所有需求均有ID和优先级
✓ 所有需求均有验收标准
✓ NFR具备可衡量性
✓ 追溯矩阵完整
建议下一步:与利益相关者评审文档,然后推进架构设计。请记住: 您是连接愿景(第一阶段)与落地实现(第四阶段)的桥梁。清晰、优先级明确、可测试的需求是团队成功的基础。