迭代决策全流程 🤖

核心原则

1. 优先级是资源分配的量化表达：Backlog排序的本质是决定资源投向，每个排序变化都意味着资源的重新分配
2. 关联即杠杆：识别需求间的依赖和协同关系，关联需求一起做比分散做效率高3倍
3. 技术债务是隐形成本，必须显性化：技术债务不进Backlog就永远不会被还，必须在优先级评分中占权重
4. 优先级调整不是重新排序，而是重新分配资源：每次调整都意味着已有承诺的打破，必须评估连锁影响
5. 变更影响评估先于调整决策：先量化影响再决定调整，避免拍脑袋调整导致更大的混乱
6. 风险评估是调整的护栏：每次调整必须附带风险评估，确保不会因调整引入更大的问题
7. 复盘的目的是改进不是追责：复盘必须建立心理安全感，否则团队只会报喜不报忧
8. 数据驱动归因，而非主观印象：用指标数据验证"感觉"，避免印象偏差掩盖真实问题
9. 改进建议必须可追踪可验证：每条改进建议都要有责任人和验证标准，否则复盘就是走过场

触发条件

交互模式

🤖→👤 AI建议人类审批

输入

执行步骤

Step 1: Backlog整理（from iteration-backlog）

目标：整理与优化问题Backlog，完成关联分析与重组

#### 1.1 问题优先级评估

评估模型：

Priority Score = Business Impact × User Value × Urgency × Effort Adjusted

评分维度：

评分公式：

priority_score:
  business_impact:
    score: 0-10
    factors:
      revenue_impact: {value}
      brand_impact: {value}
      strategic_value: {value}
  user_value:
    score: 0-10
    factors:
      request_frequency: {count}
      pain_point_intensity: {value}
  urgency:
    score: 0-10
    factors:
      alert_impact: {value}
      competitive_threat: {value}
      compliance_requirement: {value}
  effort_adjusted:
    score: 0-10
    factors:
      resource_requirement: {story_points}
      dependencies: {count}
      technical_risk: {value}
  final_score: {weighted_sum}

#### 1.2 技术债务影响分析

影响类型：

债务分类：

technical_debt_impact:
  - debt_id: {id}
    category: code_quality | performance | security | architecture
    severity: critical | high | medium | low
    affected_systems: [{system}]
    metrics:
      development_overhead: {percentage}
      defect_rate_impact: {percentage}
      performance_impact: {percentage}
    affected_backlog_items: [{item_id}]
    interest_accrued: {story_points_per_sprint}

#### 1.3 关联分析

关联类型：

关联输出：

linked_issues:
  - item_id: {id}
    dependencies:
      - depends_on: {item_id}
        type: hard | soft
        reason: {description}
    synergies:
      - related_to: {item_id}
        reason: "一起实现价值最大化"
    technical_debt_blockers:
      - debt_id: {id}
        impact: "导致开发效率降低 20%"

#### 1.4 Backlog 重组

重组策略：

重组建议格式：

reorganization_suggestions:
  - action: prioritize | combine | postpone | split | reorder
    target:
      item_id: {id}
      current_position: {position}
      suggested_position: {position}
    reason: {description}
    impact:
      effort_saved: {story_points}
      risk_reduced: {percentage}
      value_delivered: {description}

Step 2: 优先级评估（from iteration-prioritization）

目标：基于数据评估问题优先级，生成调整方案与风险评估

#### 2.1 变更影响评估

影响维度：

影响计算：

impact_assessment:
  trigger_event:
    type: monitoring_alert | user_feedback | strategic_change | resource_change
    severity: P0 | P1 | P2
    description: {description}
  scope_impact:
    affected_items:
      - item_id: {id}
        current_status: in_progress | planned
        impact_type: replace | postpone | remove | add
        story_points_affected: {points}
    total_story_points: {points}
    percentage_of_iteration: {percentage}
  schedule_impact:
    original_end_date: {date}
    estimated_end_date: {date}
    days_delayed: {days}
  quality_impact:
    risk_level: high | medium | low
    areas_affected: [{area}]
  team_impact:
    workload_change: {percentage}
    context_switches: {count}
  business_impact:
    kpis_affected: [{kpi_name}]
    impact_assessment: {description}

#### 2.2 调整方案生成

方案类型：

方案生成：

adjustment_options:
  - option_id: OPT-001
    option_type: insert | replace | postpone | split
    title: {option_title}
    description: {description}
    changes:
      items_to_add:
        - item_id: {id}
          story_points: {points}
          source: {trigger_event}
      items_to_remove:
        - item_id: {id}
          story_points: {points}
          reason: {reason}
      items_to_modify:
        - item_id: {id}
          modification: {description}
    tradeoffs:
      scope: "放弃 {feature}"
      schedule: "延期 {X} 天"
      quality: "引入 {risk}"
      business: "影响 {KPI}"
    risks:
      - risk: {description}
        likelihood: high | medium | low
        mitigation: {description}
    recommendation_score: {score}
  - option_id: OPT-002
    # 第二个方案...

#### 2.3 风险评估

风险矩阵：

风险评估输出：

risk_assessment:
  option_id: OPT-001
  overall_risk_score: {score}
  risk_breakdown:
    technical_risk:
      score: 3
      concerns: [{concern}]
    schedule_risk:
      score: 4
      concerns: [{concern}]
    quality_risk:
      score: 2
      concerns: [{concern}]
    communication_risk:
      score: 3
      concerns: [{concern}]
  mitigation_plan:
    - risk: {description}
      strategy: avoid | mitigate | transfer | accept
      action: {description}

#### 2.4 沟通草案

干系人：

• 团队成员
• 产品负责人
• 利益相关者
• 客户（如适用）

沟通模板：

communication_draft:
  recipients:
    - team_members
    - product_owner
    - stakeholders
  subject: "迭代 {sprint_name} 变更通知"
  sections:
    change_summary:
      content: "{summary_text}"
    impact:
      content: "{impact_text}"
    decisions:
      content: "{decisions_text}"
    timeline:
      content: "{timeline_text}"
    questions_contact:
      content: "{contact_info}"

Step 3: 迭代回顾（from iteration-retrospective）

目标：回顾迭代执行效果，总结经验与改进点

#### 3.1 数据收集

数据源：

数据收集范围：

data_collection:
  iteration_id: {id}
  period:
    start: {ISO8601}
    end: {ISO8601}
  delivery_data:
    planned_items: {count}
    completed_items: {count}
    planned_story_points: {points}
    completed_story_points: {points}
    carry_over_items: {count}
  quality_data:
    bugs_found: {count}
    bugs_fixed: {count}
    bug_leakage_rate: {percentage}
    code_coverage: {percentage}
    deployment_frequency: {count}
  team_feedback:
    retro_items: {count}
    top_votes: {items}
    sentiment: positive | neutral | negative
  monitoring_data:
    availability: {percentage}
    performance_change: {delta}
    incidents: {count}

#### 3.2 多维度分析

##### 3.2.1 交付分析

指标：

分析：

delivery_analysis:
  completion_rate: {percentage}
  velocity_actual: {points}
  velocity_planned: {points}
  velocity_accuracy: {ratio}
  change_rate: {percentage}
  carry_over_items:
    - item_id: {id}
      reason: {reason}
  assessment: good | acceptable | needs_improvement

##### 3.2.2 质量分析

指标：

分析：

quality_analysis:
  bug_density: {ratio}
  bug_leakage_rate: {percentage}
  code_coverage: {percentage}
  deployment_stability:
    success_rate: {percentage}
    rollbacks: {count}
  assessment: good | acceptable | needs_improvement

##### 3.2.3 协作分析

指标：

分析：

collaboration_analysis:
  team_satisfaction_score: {score}
  top_positives:
    - {item}
  top_pain_points:
    - {item}
  cross_team_collaboration:
    score: {score}
    issues: [{issue}]
  assessment: good | acceptable | needs_improvement

##### 3.2.4 效率分析

指标：

分析：

efficiency_analysis:
  team_throughput: {points_per_day}
  context_switches:
    average: {count}
    total: {count}
  blocked_time_percentage: {percentage}
  dependency_issues:
    - issue: {description}
      duration: {days}
      impact: {story_points_lost}
  assessment: good | acceptable | needs_improvement

#### 3.3 问题识别

问题分类：

问题识别输出：

problem_identification:
  - problem_id: PRB-001
    category: process | technical | collaboration | environment
    severity: P1 | P2 | P3
    description: {description}
    evidence:
      - metric: {name}
        value: {value}
        baseline: {baseline}
        deviation: {deviation}
      - feedback: "{quote}"
    root_cause_analysis:
      - question: "为什么 {problem}？"
        answer: "{cause}"
    impact:
      items_affected: {count}
      effort_lost: {story_points}
      quality_impact: {description}

#### 3.4 改进建议

建议格式：

improvement_suggestions:
  - suggestion_id: IMP-001
    problem_id: PRB-001
    category: process | technical | collaboration | environment
    title: {title}
    description: {description}
    expected_outcome:
      metric_improvement:
        - metric: {name}
          current: {value}
          target: {value}
    action_items:
      - action: {description}
        owner: {role}
        deadline: {date}
    effort_required:
      story_points: {points}
      time_estimate: {days}
    priority: P1 | P2 | P3
    recommendation_score: {score}

输出

输出文件路径：output/pm-monitoring/iteration-decision/ 输出Schema：

{
  "type": "object",
  "required": ["prioritized_items", "backlog_size", "trigger_id", "impact_assessment", "options", "iteration_id", "summary", "metrics_analysis"],
  "properties": {
    "generated_at": {"type": "string", "description": "生成时间"},
    "backlog_size": {"type": "object", "description": "Backlog规模，包含总条目数和总故事点"},
    "prioritized_items": {"type": "array", "description": "排序后的需求列表，包含评分、影响和关联"},
    "technical_debt_priority": {"type": "array", "description": "技术债务优先级列表，包含利息和优先级"},
    "reorganization_summary": {"type": "object", "description": "重组建议汇总，包含提升/合并/推迟/拆分数量"},
    "trigger_id": {"type": "string", "description": "触发事件ID"},
    "trigger_type": {"type": "string", "description": "触发类型：monitoring_alert/feedback/strategy_change"},
    "impact_assessment": {"type": "object", "description": "影响评估，包含范围/进度/质量影响"},
    "recommended_option": {"type": "string", "description": "推荐方案ID"},
    "options": {"type": "array", "description": "可选方案列表，包含类型、评分和权衡"},
    "needs_human_decision": {"type": "boolean", "description": "是否需要人工决策"},
    "iteration_id": {"type": "string", "description": "迭代ID"},
    "period": {"type": "object", "description": "迭代周期，包含起止时间"},
    "summary": {"type": "object", "description": "迭代总结，包含完成率、质量状态和评分"},
    "metrics_analysis": {"type": "object", "description": "指标分析，包含交付/质量/协作/效率四维度"},
    "problem_identification": {"type": "object", "description": "问题识别，包含总数和P1/P2计数"},
    "improvement_suggestions": {"type": "object", "description": "改进建议，包含总数和高优先级计数"}
  }
}

├── iteration-decision.json
├── iteration-decision.md
├── backlog/
│   ├── {iteration_id}/
│   │   ├── prioritized_items.yaml
│   │   ├── linked_issues.yaml
│   │   ├── technical_debt_impact.yaml
│   │   └── reorganization_suggestions.md
│   └── latest/
│       └── backlog_recommendation.md
├── prioritization/
│   ├── {trigger_id}/
│   │   ├── impact_assessment.yaml
│   │   ├── adjustment_options.yaml
│   │   ├── risk_assessment.yaml
│   │   ├── communication_draft.md
│   │   └── needs_human_decision: true | false
│   └── latest/
│       └── adjustment_recommendation.md
└── retrospective/
    ├── {iteration_id}/
    │   ├── summary.md
    │   ├── metrics_analysis.yaml
    │   ├── problem_identification.yaml
    │   └── improvement_suggestions.yaml
    └── latest/
        └── retrospective_report.md

决策规则

质量检查

• [ ] 优先级评分覆盖率 100%
• [ ] 关联关系识别完整
• [ ] 技术债务影响评估准确
• [ ] 重组建议可执行
• [ ] Sprint 容量匹配
• [ ] 无关键依赖遗漏
• [ ] 变更影响评估覆盖率 100%
• [ ] 调整方案数量 ≥ 2
• [ ] 风险评估完整性
• [ ] 沟通草案覆盖所有干系人
• [ ] 决策标记准确性
• [ ] 方案可执行性 ≥ 80%
• [ ] 数据收集完整率 ≥ 95%
• [ ] 分析覆盖所有四个维度
• [ ] 问题识别准确率 ≥ 80%
• [ ] 建议可执行率 ≥ 75%
• [ ] 改进建议有明确负责人
• [ ] 与上一迭代对比分析完整

降级策略

上游文件缺失降级方案

数据获取说明

当上游文件缺失时，通过以下方式获取必要数据：

1. 需求池缺失：请用户提供当前需求列表，包含需求标题、简述和类型（功能/Bug/优化），AI将基于提供的信息进行优先级评分和排序
2. 技术债务缺失：跳过债务影响分析步骤，优先级评分公式中移除债务相关权重，建议后续补充债务清单以完善排序
3. 监控告警缺失：跳过告警关联分析，无法自动提升紧急项优先级，建议用户手动标注紧急项
4. 用户反馈缺失：用户价值维度评分将基于需求描述推断，输出中标注该维度置信度低，建议人工确认
5. 当前迭代计划缺失：请用户描述需要调整的原因（如"支付功能出现线上问题需要加塞"），AI将基于描述生成调整方案，包含加塞/替换/推迟等选项
6. 触发事件缺失：请用户说明调整触发原因和紧迫程度，AI将据此进行影响评估和方案生成
7. 资源约束缺失：方案生成时跳过容量匹配验证，所有方案标注"需人工确认团队容量是否支持"，建议后续补充资源数据
8. 迭代完成情况缺失：请用户提供迭代完成情况，包括：计划完成的故事点/实际完成的故事点、未完成项及原因、需求变更情况，AI将基于提供的数据生成复盘报告
9. 质量指标缺失：请用户提供关键质量数据（Bug数量、严重程度分布、返工次数），AI将据此进行质量维度分析
10. 团队反馈缺失：跳过协作分析维度，复盘中标注该维度数据缺失，建议后续通过Retro会议补充
11. 监控数据缺失：跳过稳定性分析，复盘中标注缺少稳定性数据，建议从监控系统导出补充

输出校验规则

上游变更响应

上游变更影响表

下游通知机制表

版本历史

• v3.0: 合并 iteration-backlog + iteration-prioritization + iteration-retrospective