Back to Details

feature-planning

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Feature Planning Skill

功能规划Skill

Answers the question: In what order should tasks be executed? Which can run in parallel? What blocks what?
This skill focuses on task sequencing and dependency analysis, transforming a flat task list into a logical execution flow.
解决以下问题：任务应按什么顺序执行？哪些任务可并行执行？哪些任务会阻塞其他任务？
此Skill专注于任务排序与依赖分析，将扁平化的任务列表转化为逻辑化的执行流程。

When to Use

适用场景

Use this skill when you have a feature breakdown document and need to:
  • Determine the correct order to execute tasks
  • Identify which tasks block other tasks
  • Find opportunities for parallelization
  • Create a sequenced task list for execution
  • Understand dependencies and integration points
Key indicator: You're asking "In what order should these tasks be done?" and "Which can happen in parallel?"
Do NOT use this skill if:
  • You only have a feature spec/idea (use feature-breakdown first to identify all tasks)
  • You need calendar dates and deadlines (not the focus of this skill)
当你拥有一份feature breakdown文档且需要完成以下事项时，可使用此Skill：
  • 确定任务的正确执行顺序
  • 识别哪些任务会阻塞其他任务
  • 寻找并行执行的机会
  • 生成用于执行的有序任务列表
  • 了解依赖关系与集成点
关键指标：你正在询问“这些任务应按什么顺序执行？”以及“哪些任务可并行执行？”
请勿使用此Skill的场景：
  • 如果你仅拥有功能规格说明/想法（请先使用feature-breakdown来识别所有任务）
  • 如果你需要日历日期与截止期限（这并非此Skill的关注重点）

Prerequisites Check

前置条件检查

⚠️ CRITICAL: This skill requires a feature breakdown document as input. Do NOT proceed without it.
Before using this skill:
  1. Verify you have completed 
    feature-breakdown
    first
    • If you only have a feature spec or idea → STOP and use 
      feature-breakdown
      skill first
    • If you have a feature breakdown document → Continue below
  2. Expected inputs from feature-breakdown:
    • docs/features/[feature-name]-breakdown.md
      file exists
    • Contains all 8 required sections
    • Has a complete task list with dependencies and acceptance criteria
If you don't have a breakdown document:
  • Load the 
    feature-breakdown
    skill first
  • Follow its workflow to decompose your feature spec into tasks
  • Once you have 
    docs/features/[feature-name]/breakdown.md
    , return here
⚠️ 重要提示：此Skill需要以feature breakdown文档作为输入，无此文档请勿继续。
使用此Skill前：
  1. 确认你已先完成
    feature-breakdown
    • 如果你仅拥有功能规格说明或想法 → 停止操作，先使用
      feature-breakdown
      Skill
    • 如果你已有feature breakdown文档 → 继续以下步骤
  2. feature-breakdown的预期输入：
    • 存在
      docs/features/[feature-name]-breakdown.md
      文件
    • 包含全部8个必填章节
    • 拥有包含依赖关系与验收标准的完整任务列表
如果你没有拆解文档：
  • 先加载
    feature-breakdown
    Skill
  • 按照其工作流将你的功能规格说明拆解为任务
  • 当你拥有
    docs/features/[feature-name]/breakdown.md
    后，返回此处

Inputs

输入项

  • Feature breakdown document (required): 
    docs/features/[feature-name]/breakdown.md
  • Feature breakdown文档（必填）：
    docs/features/[feature-name]/breakdown.md

Outputs

输出项

MANDATORY FILE ORGANIZATION: All feature files must be in 
docs/features/<feature-name>/
subdirectory.
When this skill completes, it creates:
  1. Sequenced Task List (
    docs/features/[feature-name]/implementation-sequence.md
    )
    • All tasks organized in optimal execution order
    • Clear dependency relationships
    • Parallelization opportunities identified
    • Ready for AI agents to execute
    • Example: 
      docs/features/user-authentication/implementation-sequence.md
强制文件组织规则：所有功能相关文件必须存放于
docs/features/<feature-name>/
子目录下。
当此Skill执行完成后，将生成：
  1. 有序任务列表（
    docs/features/[feature-name]/implementation-sequence.md
    ）
    • 所有任务按最优执行顺序排列
    • 清晰的依赖关系
    • 已识别并行执行机会
    • 可供AI Agent直接执行
    • 示例：
      docs/features/user-authentication/implementation-sequence.md

Workflow Overview

工作流概述

The feature planning process transforms a breakdown into an ordered execution sequence:
Feature Breakdown Input
    ↓
Extract Tasks & Dependencies
    ↓
Build Dependency Graph
    ↓
Identify Execution Sequence
    ↓
Mark Parallel Opportunities
    ↓
Create Sequenced Task List
    ↓
Ready for Execution
功能规划流程将拆解内容转化为有序执行序列：
Feature Breakdown输入
    ↓
提取任务与依赖关系
    ↓
构建依赖关系图
    ↓
确定执行顺序
    ↓
标记并行执行机会
    ↓
生成有序任务列表
    ↓
准备执行

Core Workflow

核心工作流

Phase 1: Extract and Validate Breakdown

阶段1：提取并验证拆解内容

Input: Feature breakdown document
  1. Read breakdown structure:
    • Extract all tasks with their IDs and descriptions
    • Document all dependencies (which tasks block which)
    • Note component mappings
    • List acceptance criteria for each task
  2. Validate task quality:
    • Confirm each task has acceptance criteria
    • Verify dependencies are explicitly stated
    • Identify any circular dependencies
    • Check for gaps or missing tasks
输入：Feature breakdown文档
  1. 读取拆解结构：
    • 提取所有带ID与描述的任务
    • 记录所有依赖关系（哪些任务阻塞哪些任务）
    • 记录组件映射关系
    • 列出每个任务的验收标准
  2. 验证任务质量：
    • 确认每个任务都有验收标准
    • 验证依赖关系已明确说明
    • 识别是否存在循环依赖
    • 检查是否存在任务缺口或遗漏任务

Phase 2: Build Dependency Graph

阶段2：构建依赖关系图

  1. Map all task dependencies:
    • Create visual representation of what blocks what
    • Identify sequential tasks (must complete in order)
    • Identify parallel tasks (can run simultaneously)
    • Flag any circular dependencies or issues
  2. Identify critical path (longest dependency chain):
    • Calculate total length of each dependency chain
    • Mark the longest chain as critical path
    • Note tasks on critical path that cannot slip
  3. Group related work:
    • Group tasks by component
    • Identify natural groupings for batching
    • Recognize integration points requiring coordination
  1. 映射所有任务依赖关系：
    • 创建任务阻塞关系的可视化表示
    • 识别顺序任务（必须按顺序完成）
    • 识别并行任务（可同时执行）
    • 标记任何循环依赖或问题
  2. 识别关键路径（最长的依赖链）：
    • 计算每条依赖链的总长度
    • 将最长的链标记为关键路径
    • 记录关键路径上不可延误的任务
  3. 分组相关工作：
    • 按组件对任务进行分组
    • 识别适合批量处理的自然分组
    • 识别需要协调的集成点

Phase 3: Determine Execution Sequence

阶段3：确定执行顺序

  1. Order tasks respecting dependencies:
    • Start with tasks that have no dependencies (can start immediately)
    • Place dependent tasks after their prerequisites
    • Group parallel tasks together
    • Organize by logical component flow
  2. Identify parallelization opportunities:
    • Mark which tasks CAN run simultaneously
    • Group parallel tasks for batch execution
    • Document why tasks can/cannot run in parallel
  1. 遵循依赖关系排序任务：
    • 从无依赖的任务开始（可立即启动）
    • 将依赖任务置于其前置任务之后
    • 将并行任务分组
    • 按逻辑组件流组织任务
  2. 识别并行执行机会：
    • 标记可同时执行的任务
    • 将并行任务分组以便批量执行
    • 记录任务可/不可并行执行的原因

Phase 4: Create Sequenced Task List

阶段4：生成有序任务列表

Generate the sequenced task list with this structure:
Sequenced Task Structure:
markdown
undefined
按以下结构生成有序任务列表：
有序任务结构：
markdown
undefined

Task [ID]: [Task Title]

任务[ID]：[任务标题]

Component: [Which component(s)]
Depends On: [List prerequisite task IDs, or "None"]
Parallel With: [Other task IDs that can run simultaneously, or "None"]
Description: [What needs to be done]
Acceptance Criteria:
  • Criterion 1
  • Criterion 2
  • Criterion 3
Integration Points: [What does this task interface with?]
Risks: [Any technical risks or complexity?]

**Organization principles**:

1. **Order tasks sequentially** - respect dependencies
2. **Group parallel tasks** - related parallel work together
3. **Mark critical path** - indicate which tasks determine overall completion
4. **Note integration points** - show what requires coordination
5. **Include all details** - make each task self-contained and clear
组件：[涉及哪些组件]
依赖于：[前置任务ID列表，或“无”]
可并行任务：[可同时执行的其他任务ID，或“无”]
描述：[需要完成的工作]
验收标准：
  • 标准1
  • 标准2
  • 标准3
集成点：[此任务与哪些内容交互？]
风险：[是否存在技术风险或复杂度？]

**组织原则**：

1. **按顺序排列任务** - 遵循依赖关系
2. **分组并行任务** - 将相关并行任务放在一起
3. **标记关键路径** - 指明哪些任务决定整体完成时间
4. **记录集成点** - 显示哪些内容需要协调
5. **包含所有细节** - 使每个任务独立且清晰

Phase 5: Identify Batch Groupings

阶段5：识别批量分组

For single developer + AI agent execution:
  1. Group tasks into logical batches (1-3 tasks per batch):
    • Batch 1: Foundation/Setup tasks
    • Batch 2: Component A implementation
    • Batch 3: Component B implementation
    • Batch 4: Integration & Testing
    • Etc.
  2. Each batch should:
    • Be executable without waiting for other work
    • Have clear success criteria
    • Be able to complete before next batch starts
    • Have manageable scope for one agent session
针对单个开发者+AI Agent的执行场景：
  1. 将任务分组为逻辑批量（每批1-3个任务）：
    • 批量1：基础/设置任务
    • 批量2：组件A实现
    • 批量3：组件B实现
    • 批量4：集成与测试
    • 以此类推
  2. 每批任务应满足：
    • 无需等待其他工作即可执行
    • 拥有清晰的成功标准
    • 能够在启动下一批任务前完成
    • 单个Agent会话可处理的合理范围

Output Format

输出格式

Create a single file: 
docs/features/[feature-name]/implementation-sequence.md
  • Directory: Must use feature-specific subdirectory: 
    docs/features/[feature-name]/
  • Filename: 
    implementation-sequence.md
    (clear, representative name)
  • Example: 
    docs/features/user-authentication/implementation-sequence.md
  • MANDATORY: All feature files in same directory to maintain organization
markdown
undefined
生成单个文件：
docs/features/[feature-name]/implementation-sequence.md
  • 目录：必须使用功能专属子目录：
    docs/features/[feature-name]/
  • 文件名：
    implementation-sequence.md
    （清晰且具有代表性的名称）
  • 示例：
    docs/features/user-authentication/implementation-sequence.md
  • 强制要求：所有功能相关文件需存放在同一目录以保持组织性
markdown
undefined

Execution Sequence: [Feature Name]

执行顺序：[功能名称]

Overview

概述

Total Tasks: [N] Critical Path Length: [N] tasks Parallel Opportunities: [N] groups of parallel tasks
总任务数：[N] 关键路径长度：[N]个任务 并行执行机会：[N]组并行任务

Dependency Graph

依赖关系图

[Visual ASCII representation or text description of dependencies]
[依赖关系的ASCII可视化表示或文字描述]

Sequenced Task List

有序任务列表

[All tasks in order, following structure above]
[所有任务按上述结构排序]

Batch Groupings (For Agent Execution)

批量分组（供Agent执行）

Batch 1: [Batch Name]

批量1：[批量名称]

  • Task [ID]: [Title]
  • Task [ID]: [Title]
  • Can execute in parallel: [Yes/No]
  • Prerequisites: [None or list]
  • 任务[ID]：[标题]
  • 任务[ID]：[标题]
  • 可并行执行：[是/否]
  • 前置条件：[无或列表]

Batch 2: [Batch Name]

批量2：[批量名称]

[etc]
[以此类推]

Critical Path

关键路径

Tasks that determine overall completion time:
  • Task [ID] → Task [ID] → Task [ID]
  • Total length: [N] tasks
  • Cannot slip without delaying project
决定整体完成时间的任务：
  • 任务[ID] → 任务[ID] → 任务[ID]
  • 总长度：[N]个任务
  • 任何延误都会导致整个功能延期

Integration Points

集成点

Key coordination requirements:
  • Task A output becomes Task B input
  • [etc]
关键协调要求：
  • 任务A的输出作为任务B的输入
  • [以此类推]

Next Steps

下一步操作

Execute batches sequentially, using this sequence as the source of truth.
undefined
以此序列为依据，按顺序执行各批量任务。
undefined

Guidelines

指导原则

Dependency Rules

依赖规则

  • Hard Dependency: Task B cannot start until Task A is 100% complete
  • Soft Dependency: Task B can start during Task A but needs early completion milestone
  • No Dependency: Tasks are completely independent
  • 强依赖：任务B需在任务A100%完成后才能启动
  • 弱依赖：任务B可在任务A执行期间启动，但需要任务A达到早期完成里程碑
  • 无依赖：任务完全独立

Parallelization

并行执行

Tasks can run in parallel if:
  • They have no dependencies on each other
  • They work on different components
  • They don't share resources
Tasks must run sequentially if:
  • One's output feeds into the other's input
  • They modify the same file/component
  • One sets up infrastructure the other needs
满足以下条件的任务可并行执行：
  • 彼此无依赖关系
  • 处理不同组件
  • 不共享资源
必须按顺序执行的任务：
  • 一个任务的输出作为另一个任务的输入
  • 修改同一文件/组件
  • 一个任务为另一个任务搭建所需基础设施

Critical Path

关键路径

The critical path is what determines overall completion. If ANY task on the critical path slips, the entire feature slips.
Identify critical path by:
  1. Calculate total length of each dependency chain
  2. Find the longest chain
  3. These are the critical tasks
关键路径决定了整体完成时间。关键路径上的任何任务延误都会导致整个功能延期。
识别关键路径的方法：
  1. 计算每条依赖链的总长度
  2. 找到最长的链
  3. 该链上的任务即为关键任务

Common Pitfalls to Avoid

需避免的常见陷阱

❌ Ignoring dependencies: Ordering tasks that block each other
❌ Assuming parallel work: Tasks that look independent but share dependencies
❌ Missing integration points: Not recognizing when one task needs output from another
❌ Circular dependencies: Task A depends on B, B depends on A
❌ Vague grouping: Batches that are too large to execute in one session
❌ 忽略依赖关系：将互相阻塞的任务排序
❌ 假设可并行执行：看似独立但实际存在共享依赖的任务
❌ 遗漏集成点：未识别到某个任务需要另一个任务的输出
❌ 循环依赖：任务A依赖任务B，任务B依赖任务A
❌ 模糊分组：批量任务范围过大，无法在单个会话中执行

See Also

另请参阅

For reference materials, see the included reference documents:
  • sequencing-guide.md
    : How to analyze and sequence dependencies
  • batch-organization.md
    : How to group tasks for execution
参考资料，请查看附带的参考文档：
  • sequencing-guide.md
    ：如何分析与排序依赖关系
  • batch-organization.md
    ：如何为执行分组任务