关于 shape-up
shape-up 是一项将 Basecamp Shape Up 方法用于产品规划与开发的技能,适合需要控制项目范围、安排周期并提升团队自主性的团队。它强调先塑造方案、按固定时间投入,并在周期内灵活调整范围。
主要功能
- 01塑造工作方案与问题定义
- 02以投入时长代替工期估算
- 03规划六周开发与两周冷却周期
- 04通过下注表确定周期项目
- 05区分必需项与可延后范围
适用场景
- 01为功能构建明确范围与方案
- 02规划下一开发周期的项目选择
- 03进度落后时削减非核心工作
安装
复制下面的命令,在终端中运行。
npx skills add https:/ / github.com/ guia-matthieu/ clawfu-skills --skill shape-up安装后在 Claude Code 中输入/shape-up确认已加载。
详细介绍
Shape Up
摆脱构建陷阱和无休止的待办事项列表。运用 Basecamp 的方法论,以固定时间、可变范围的方式,在 6 周周期内交付有意义的工作。
何时使用此技能
- 产品规划:取代无休止的待办事项列表
- 功能开发:设定明确的时间边界
- 团队自主性:希望团队能自主开展工作
- 范围管理:项目容易不断膨胀时
- 初创企业开发:资源有限时
- 代理商/咨询项目:时间线固定时
方法论基础
| 方面 | 详情 |
|---|---|
| 来源 | Ryan Singer - Shape Up(2019),由 Basecamp 开发 |
| 核心原则 | “时间固定,范围可变。考虑投入意愿,而非估算。先塑形,再构建。” |
| 为何重要 | 传统方法要么事无巨细地管理(瀑布式),要么留有太多不确定性(缺乏方向的敏捷冲刺)。Shape Up 既为团队指明方向,也赋予团队自主权。 |
Claude 的职责与由你决定的事项
| Claude 的职责 | 由你决定 |
|---|---|
| 梳理制作工作流 | 最终创意方向 |
| 提出技术方案 | 设备和工具选择 |
| 创建模板和检查清单 | 质量标准 |
| 确定最佳实践 | 品牌/风格决策 |
| 生成脚本大纲 | 最终脚本审批 |
此技能的用途
- 介绍塑形 - 在恰当的抽象层次定义工作
- 以投入意愿取代估算 - 这项工作值得投入多少时间?
- 启用周期 - 6 周专注工作,2 周冷却期
- 赋能团队 - 在边界内自主行动
- 提供下注表 - 基于原则确定优先级
- 动态管理范围 - 区分必需项与锦上添花项
使用方法
塑造功能构想
I want to shape this feature idea: [description]
Apply Shape Up methodology to define it at the right level.
Appetite: [2 weeks / 6 weeks]
规划周期
We have these potential projects for the next cycle:
[List of ideas]
Help me run a betting table to decide what to build.
在构建过程中管理范围
We're in week 3 of a 6-week cycle building [feature].
We're running behind. Help me apply Shape Up scope hammering.
说明
第 1 步:理解 Shape Up 原则
## The Shape Up Philosophy
### Fixed Time, Variable Scope
**Traditional approach:**
"How long will this take?" → Estimate → Build → Deadline slips
**Shape Up approach:**
"How much time is this worth?" → Appetite → Shape to fit → Ship on time
**The mindset shift:**
Instead of estimating how long a feature will take,
decide how much time you're willing to spend.
Then shape the work to fit that time.
### Appetite, Not Estimates
**Appetite:** How much time is this problem WORTH solving?
- Small batch: 2 weeks or less
- Big batch: 6 weeks max
**Key insight:**
A feature can be built in 2 weeks OR 6 months.
The question is: What version fits your appetite?
**Example:**
"Auto-complete for search"
- 6-month version: ML-powered, personalized, learns preferences
- 6-week version: Pre-populated common searches, basic matching
- 2-week version: Static list of top searches
All solve the problem. Choose based on appetite.
### Shaping vs. Building
**Shaping (Senior people):**
- Define the problem
- Set boundaries
- Identify risks
- Rough solution direction
- Leave room for builder creativity
**Building (Teams):**
- Detailed implementation
- Technical decisions
- UX specifics
- Scope management within boundaries
第 2 步:塑形流程
## How to Shape Work
### Step 1: Set the Appetite
Before anything else, decide:
- Is this a **small batch** (2 weeks) or **big batch** (6 weeks)?
- Is this worth doing at all at this appetite?
**Questions to ask:**
- What problem are we solving?
- How painful is this problem?
- What's the opportunity cost of not doing it?
- What's the opportunity cost of spending more time on it?
### Step 2: Narrow the Problem
Don't shape "improve search."
Shape "help new users find their first project template."
**Narrowing technique:**
1. Start with the raw idea
2. Ask: Who specifically has this problem?
3. Ask: In what specific situation?
4. Ask: What's the minimum viable solution?
### Step 3: Rough Out the Solution
**Fat marker sketches:**
Draw the solution with a thick marker (no detail).
You're defining spaces and flows, not buttons and fields.
**Breadboarding:**
For flows, use words not wireframes:
[搜索框] → [结果页面] → [模板详情] ↓ [无结果] → [建议类别]
**Key principle:**
Leave room for the builders to be creative.
Define WHAT, not exactly HOW.
### Step 4: Identify Risks and Rabbit Holes
**Rabbit holes:** Technical or design problems that could explode in scope.
**For each potential rabbit hole:**
- Name it
- Decide: Solve it in shaping? Or declare it out of scope?
- Document the boundary
**Example:**
"If we build template search, what about user-generated templates?"
Decision: Out of scope. Only show official templates.
### Step 5: Write the Pitch
**Pitch elements:**
1. **Problem:** What are we solving?
2. **Appetite:** How long is this worth?
3. **Solution:** Fat marker sketch / breadboard
4. **Rabbit holes:** What we're explicitly NOT doing
5. **No-gos:** Boundaries and constraints
第 3 步:周期
## Six-Week Cycles
### The Rhythm
**6 weeks building:**
- Long enough for meaningful work
- Short enough to maintain urgency
- Teams own their projects completely
**2 weeks cooldown:**
- Bug fixes
- Technical debt
- Exploration
- Shaping for next cycle
- Recovery
### Why 6 Weeks?
**Shorter (2-week sprints):**
- Not enough time for real progress
- Constant planning overhead
- Work gets chopped up artificially
**Longer (quarters):**
- Deadlines feel far away
- Scope creeps
- No urgency until the end
**6 weeks:**
- Urgent from day one
- Room to figure things out
- Clean endpoint
### Team Structure
**Small teams:**
- 1-2 designers + 1-3 programmers
- Self-managed during the cycle
- No daily standups with managers
- Check-ins when THEY need help
**Circuit breaker:**
If work isn't done at 6 weeks, it doesn't automatically continue.
It goes back to the betting table. Maybe it gets another cycle.
Maybe it doesn't.
### What Teams Do in a Cycle
**Week 1-2: Figure it out**
- Understand the shaped work
- Spike on unknowns
- Get oriented
- Early integration
**Week 3-4: Build the core**
- Make vertical slices
- Connect the pieces
- Working software early
**Week 5-6: Polish and ship**
- Cut scope if needed
- Must-haves only
- Ship by end of cycle
第 4 步:下注表
## Choosing What to Build
### The Betting Table
**Who:** Senior people who can make commitments
**When:** During cooldown, before next cycle
**Input:** Shaped pitches
**Output:** Cycle bets
### The Process
**1. Review pitches**
Each pitch should be complete:
- Clear problem
- Shaped solution
- Identified risks
- Appetite set
**2. Consider each bet**
For each pitch, ask:
- Is this the right time?
- Do we have the right team?
- Are there dependencies?
- What's the opportunity cost?
**3. Make decisions**
Options:
- **Bet:** Assign to next cycle
- **Park:** Good but not now
- **Kill:** Not worth doing
**No backlog:**
If you don't bet on something, it goes away.
Good ideas come back. Bad ideas don't.
### Betting Criteria
**1. Strategic fit**
Does this support current company goals?
**2. Problem significance**
How painful is this for customers?
**3. Appetite match**
Can this actually be done in the proposed time?
**4. Team availability**
Who would work on this?
**5. Dependencies**
What else needs to be true?
### Anti-Patterns
**Carry-over:**
"We didn't finish last cycle, so we'll continue."
No. Circuit breaker. Re-evaluate. Maybe it's not worth it.
**Backlog grooming:**
"Let's go through the 200 ideas and prioritize."
No. Only consider shaped pitches. Unshaped ideas aren't real options.
**Consensus:**
"Let's vote on what to build."
No. Decision-makers decide. Not democracy.
第 5 步:管理范围
## Scope Hammering
### The Principle
Scope grows naturally. Left unchecked, projects expand to fill time.
Your job is to constantly hammer scope back to what matters.
### Must-Haves vs. Nice-to-Haves
**Must-haves:**
- Core value delivery
- Without this, the feature doesn't work
- Absolutely required for ship
**Nice-to-haves:**
- Polish
- Edge cases
- Delighters
- "While we're at it..."
**Rule:** Identify nice-to-haves early. Cut them first.
### The Scope Hammer
**When you're behind (week 3+):**
1. **List all remaining work**
2. **Classify each item:**
- Must-have for THIS version
- Nice-to-have (cut or defer)
3. **Cut aggressively:**
- What's the smallest version that still delivers value?
- What can be a follow-up?
- What can we not do at all?
### Scope Hammering Questions
- "Is this necessary for launch, or is it polish?"
- "What's the simplest version that still solves the problem?"
- "Can this be a fast-follow?"
- "What would happen if we just... didn't do this?"
- "Are we building for the 80% case or the edge case?"
### Hill Charts
**Track progress visually:**
┌─────────────────────────────────┐
│ ↗ 功能 C │
│ ↗ 功能 B │
│ ↗ 功能 A ↘ 功能 D
└─────────────────────────────────┘
弄清楚方向 实现目标
- **Uphill:** Still figuring it out (risky)
- **Downhill:** Know what to do, just executing (safe)
**Use for:**
- Early warning when something is stuck uphill
- Identifying where help is needed
- Manager visibility without micromanagement
示例
示例 1:塑造一项功能
输入:
“我们想为项目管理应用添加通知。你能用 Shape Up 帮我塑造这个功能吗?”
输出:
已塑形提案:项目通知
问题
用户只偶尔查看应用,因此会错过重要更新。 他们查看应用时,面对的是海量动态,无法判断 哪些内容真正重要。
具体情境: 项目经理离开一天后登录, 无法快速判断哪些事项需要关注。
投入意愿
6 周(大批次)
通知是基础功能,值得认真投入。 但不值得花 3 个月。我们会将范围控制在 6 周内。
解决方案
面包板图:
[Activity happens] → [Notification created] → [Notification center]
↓
[Settings: what to notify]
粗略标记草图:
┌─────────────────────────────────────┐
│ 🔔 (3) │ ← Bell in header with count
├─────────────────────────────────────┤
│ ┌─────────────────────────────────┐ │
│ │ ● @you in Project Alpha │ │ ← Unread notifications
│ │ Sarah mentioned you in a... │ │
│ │ 2 hours ago │ │
│ ├─────────────────────────────────┤ │
│ │ ○ Task completed │ │ ← Read notifications
│ │ Build specs was marked done │ │
│ │ Yesterday │ │
│ └─────────────────────────────────┘ │
└─────────────────────────────────────┘
我们要构建的内容:
- 应用内通知中心(铃铛图标、下拉菜单)
- 以下事项的通知:@提及、任务分配、截止日期
- 基本设置(可按类型开启/关闭)
偏离主题(明确不在范围内)
电子邮件通知: 这是一个完全独立的系统(送达率、退订、格式)。 不在范围内。未来可作为小批次处理。
移动推送通知: 需要开发原生应用。不在范围内。
通知分组: 将 5 条独立通知合并为“Project Alpha 有 5 条评论”。 锦上添花。如果时间紧张就砍掉。
实时更新: 使用 WebSockets 即时推送通知。锦上添花。 V1 每 60 秒轮询一次即可。
按项目自定义: “只在 Project X 有更新时通知我。”过于复杂。 先从全局设置开始。
禁止事项
- 不发送电子邮件通知
- 不发送移动推送
- 不播放通知声音
- 不提供摘要/汇总视图
- 不提供团队级设置
风险
风险 1:数据库负载 可能会创建大量通知。需要高效的查询。 缓解措施: 技术负责人应在周期的前 2 天对通知存储方案进行调研。
风险 2:UX 复杂度 设置可能很快变得复杂。 缓解措施: 先从 3 个开关开始。不提供按项目设置。
示例 2:周期中途缩减范围
输入:
“我们正在进行一个为期 6 周的周期,目前处于第 4 周,正在开发报表仪表板。进度落后了。目前计划了:(1) 5 种图表类型,只完成了 2 种,(2) 尚未开始 PDF 导出,(3) 日期筛选器基本完成,(4) 尚未开始实时刷新。帮我缩减范围。”
输出:
范围缩减:报表仪表板
当前状态(第 4 周,共 6 周)
| 项目 | 状态 | 所需时间 |
|---|---|---|
| 图表类型(计划 5 种) | 完成 2 种 | 剩余 3 种需 1.5 周 |
| 导出为 PDF | 尚未开始 | 1 周 |
| 日期筛选器 | 完成 80% | 2 天 |
| 实时刷新 | 尚未开始 | 1 周 |
剩余工作总量:约 4 周 可用时间:约 2 周
分类
必需项(交付核心价值):
- 至少 3 种图表类型(折线图、柱状图、饼图)——覆盖最常见的使用场景
- 日期筛选器——没有筛选就毫无用处
- 查看/加载报表——基本功能
锦上添花(砍掉或延期):
- 第 4 和第 5 种图表类型(表格图、面积图)
- 导出为 PDF
- 实时刷新
建议
现在砍掉:
- 实时刷新 → V2。手动刷新按钮即可。
- 第 5 种图表类型(面积图) → V2。很少使用。
- 导出为 PDF → V2。用户可以截图,或者我们在之后快速跟进。
保留:
- 日期筛选器(2 天内完成)
- 3 种图表类型(折线图、柱状图、饼图——完成剩余 1 种需 3–4 天)
- 第 4 种图表类型(表格图)——便于查看数据(3 天)
修订后的范围
第 5 周:
- 第 1–2 天:完成日期筛选器
- 第 3–4 天:构建饼图(第 3 种类型)
- 第 5 天:构建表格视图(第 4 种类型)
第 6 周:
- 第 1–3 天:优化、修复错误、处理边界情况
- 第 4 天:内部测试
- 第 5 天:发布
如何告知利益相关者
“我们将发布功能完备的仪表板,包含 4 种图表类型(折线图、柱状图、饼图、表格)和日期筛选器。导出和实时刷新将在下一个小批次周期中快速跟进。
这能为用户带来 90% 的价值。剩余功能只是优化,并非核心功能。”
周期结束后的小批次候选项
- 导出为 PDF(2 周批次)
- 实时刷新(2 周批次)
- 面积图类型(1 周附加项)
示例 3:组织一场押注讨论
输入:
“下一个周期有以下 5 个提案。如何组织一场押注讨论来决定?(1) 改进搜索,(2) 团队权限,(3) 移动应用重新设计,(4) 集成 API,(5) 改进引导流程”
输出:
押注讨论:下一个周期
前期工作:Pitch 评审
每个 pitch 都应包含:
- 清晰的问题陈述
- 投入时长(2 周或 6 周)
- 大致解决方案
- 已识别的陷阱
- 明确的禁区
评估:
| Pitch | 是否成形? | 投入时长 | 所需团队 |
|---|---|---|---|
| 改进搜索 | ✓ 是 | 6 周 | 1D + 2E |
| 团队权限 | ✓ 是 | 6 周 | 1D + 2E |
| 移动应用重新设计 | ✗ 太模糊 | ? | ? |
| 集成 API | ✓ 是 | 6 周 | 0D + 3E |
| 改进引导流程 | ✓ 是 | 2 周 | 1D + 1E |
移动应用重新设计: 尚未准备好投入。需要进一步打磨。 退回。考虑在未来周期中安排。
投入标准评估
1. 改进搜索(6 周)
| 标准 | 评分 | 备注 |
|---|---|---|
| 战略契合度 | 4/5 | 支持增长,符合用户需求 |
| 问题重要性 | 3/5 | 主要困扰高级用户 |
| 投入时长匹配度 | 4/5 | 范围明确 |
| 团队可用性 | ✓ | 团队 A 有空 |
| 依赖项 | 无 |
结论: 候选项目
2. 团队权限(6 周)
| 标准 | 评分 | 备注 |
|---|---|---|
| 战略契合度 | 5/5 | 企业交易所必需 |
| 问题重要性 | 5/5 | 阻碍销售 |
| 投入时长匹配度 | 3/5 | 范围可能扩大,需要严格控制 |
| 团队可用性 | ✓ | 团队 B 有空 |
| 依赖项 | 无 |
结论: 强力候选项目
3. 集成 API(6 周)
| 标准 | 评分 | 备注 |
|---|---|---|
| 战略契合度 | 4/5 | 拓展合作伙伴生态 |
| 问题重要性 | 3/5 | 重要但不紧急 |
| 投入时长匹配度 | 4/5 | 首期限定为只读 |
| 团队可用性 | ✓ | 团队 C 有空 |
| 依赖项 | 无 |
结论: 候选项目
4. 改进引导流程(2 周)
| 标准 | 评分 | 备注 |
|---|---|---|
| 战略契合度 | 5/5 | 直接影响激活率 |
| 问题重要性 | 4/5 | 引导流程中有 40% 的流失率 |
| 投入时长匹配度 | 5/5 | 范围小且聚焦 |
| 团队可用性 | ✓ | 可纳入任何团队的周期 |
| 依赖项 | 无 |
结论: 强力候选项目(小批次)
投入项目
可用产能:
- 2 个团队可承担 6 周项目
- 1 个团队可腾出时间增加一个 2 周项目
决定:
| 投入项目 | 团队 | 理由 |
|---|---|---|
| 团队权限 | 团队 B | 阻碍企业交易,紧迫性最高 |
| 集成 API | 团队 C | 带来战略机遇 |
| 改进引导流程 | 团队 A(第 1–2 周) | 影响大,投入小 |
| 改进搜索 | 暂缓 | 不错,但目前优先级不够高 |
不投入的项目:
- 移动应用重新设计:尚未成形,需要进一步打磨。
- 改进搜索:Pitch 不错,但时机不对。留到下个周期。
投入后的沟通
“下个周期:
- 团队 B:团队权限(6 周)
- 团队 C:API v1(6 周)
- 团队 A:改进引导流程(2 周),然后处理冷却期任务
搜索是一个很有竞争力的 pitch。我们将其暂缓,留到再下一个周期。 移动应用重新设计需要进一步打磨,之后才能投入。”
清单与模板
Pitch 模板
## Pitch: [Feature Name]
### Problem
[What problem are we solving? Who has it? When?]
### Appetite
[2 weeks / 6 weeks]
### Solution
**Breadboard:**
[Flow diagram with words]
**Fat Marker Sketch:**
[Rough visual layout - no details]
### Rabbit Holes
[What could explode in scope? How are we preventing it?]
### No-Gos
[What are we explicitly NOT building?]
### Risks
[What could go wrong? How will we mitigate?]
投入会议清单
## Betting Table: [Cycle Name]
### Before the Meeting
□ All pitches reviewed for completeness
□ Incomplete pitches sent back for shaping
□ Team availability mapped
□ Strategic priorities clear
### During the Meeting
□ Review each complete pitch
□ Assess against betting criteria
□ Discuss dependencies and timing
□ Make binary decisions (bet / don't bet)
□ Assign teams to bets
### After the Meeting
□ Communicate decisions to teams
□ Archive or park unbetted pitches
□ Schedule cycle kickoffs
□ Clear any dependencies
技能边界
此技能擅长的方面
- 规划音频制作工作流
- 提供技术指导
- 创建质量检查清单
- 提出创意方案
此技能无法做到的事
- 取代音频工程专业知识
- 做出主观的创意决策
- 直接访问或编辑音频文件
- 保证商业成功
参考资料
- Singer, Ryan.《Shape Up:Stop Running in Circles and Ship Work that Matters》(2019)
- Basecamp 方法论文档
- 37signals(Basecamp)博客文章
- Shape Up 播客访谈
Related Skills
- product-discovery - Discovery before shaping
- design-sprint - Alternative sprint format
- lean-canvas - Business model context
- first-principles - Problem definition
Skill Metadata
- Mode: cyborg
name: shape-up
category: product
subcategory: methodology
version: 1.0
author: MKTG Skills
source_expert: Ryan Singer
source_work: Shape Up
difficulty: intermediate
estimated_value: $5,000+ process consulting
tags: [product, process, Basecamp, cycles, shaping, scope, betting, development]
created: 2026-01-25
updated: 2026-01-25