Token导航 LogoToken导航

Agent Skill · 研究检索

shape-up(塑造)

shape-up 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

shape-up 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为需要联网;许可证信息还不明确,第三方扫描已通过,按安装量在研究检索类的 73,675 个 Skill 里排第 55,679。

总安装 470周安装 20GitHub Stars 35下载 165核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
unknown

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:shape-up(塑造)
来源仓库:https://github.com/petekp/claude-code-setup
仓库路径:skills/shape-up
安装命令:
npx skills add https://github.com/petekp/claude-code-setup --skill shape-up
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/petekp/claude-code-setup --skill shape-up

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

shape-up 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词快速定位候选结果时使用。适用于研究检索类任务,可结合来源仓库、安装命令和原始 README 继续核验具体用法。

通过 npx skills add 命令从指定 GitHub 仓库安装,支持多宿主环境集成。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写操作。shape-up 属于研究检索类 Skill,可作为该场景下的辅助能力补充。

来源仓库内路径:skills/shape-up

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/petekp/claude-code-setup --skill shape-up;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要需要联网,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

shape-up 怎么安装?

在终端执行 npx skills add https://github.com/petekp/claude-code-setup --skill shape-up。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

shape-up 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

shape-up 需要什么权限?

权限级别是需要联网。该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

shape-up 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

shape-up 用的人多吗?

公开数据是累计安装 470 次,近一周安装 20 次,下载 165 次,GitHub 35 星。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

shape-up 安全吗?

第三方扫描结果为通过,但这只覆盖已知的静态检查项,不能替代自己读一遍源码和安装命令。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。
  • 第三方扫描已通过,常见的高风险行为没有命中。

先想清楚的地方

  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。

不太适合

  • 需要明确授权才能用到公司项目里的场景。
  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
shape-up

塑造 · 本页

47035需要联网未明确通过

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

Shape Up

Orchestrate the full Shape Up workflow by invoking /shaping and /breadboarding at the right moments. Be proactive — after each step, explain what was accomplished and suggest the next move.

Prerequisites

Requires the /shaping and /breadboarding skills to be installed. If missing, instruct the user to install them from https://github.com/rjs/shaping-skills

Workflow Overview

Frame → Shape → Breadboard → Slice → Build
PhasePurposeSub-skillOutput
FrameCapture the "why"/shapingProblem + Outcome
ShapeSeparate problem from solution, iterate/shapingR (requirements) + selected Shape
BreadboardMap affordances and wiring/breadboardingAffordance tables + diagram
SliceCut into demoable increments/breadboardingV1..V9 vertical slices
BuildImplement slice by slice—Working software

Entry Point Detection

Determine the project type, then follow the appropriate path:

New project (empty or new directory)? → Start at Phase 1: Frame Existing project with code? → Start at Phase 0: Map Current System

Phase 0: Map Current System (Existing Projects Only)

Before shaping a change to an existing codebase, understand what exists.

  1. Ask: "What workflow or area of the system does this change affect?"
  2. Read the relevant code to understand the current implementation
  3. Invoke /breadboarding to map the existing system as CURRENT:

- Trace the affected workflow through the code - Produce affordance tables (UI + Code) for the current system - Generate a wiring diagram showing how it works today

This CURRENT breadboard becomes the baseline. Proposed shapes will modify it.

After completing: Present the CURRENT breadboard and say:

"Here's how the current system works. Now let's frame what we want to change. What problem are you trying to solve, or what outcome do you want?"

Phase 1: Frame

Capture source material and distill it into Problem and Outcome.

  1. Ask the user to describe what they want (or paste stakeholder quotes, requests, etc.)
  2. Invoke /shaping — capture the description as Source material
  3. Distill into:

- Problem — What's broken, what pain exists - Outcome — What success looks like

  1. Write to shaping.md in .claude/ (or project root for standalone projects)

After completing: Present the frame and say:

"Here's the frame. Does this capture the problem and desired outcome? Once confirmed, we'll start extracting requirements and sketching a solution."

Phase 2: Shape

Iterate on problem (R) and solution (shapes) until a shape is selected.

  1. Invoke /shaping
  2. Offer both entry points:

- Start from R — extract requirements from the frame - Start from S — if user already has a solution in mind, capture it as Shape A

  1. Run the shaping loop:
Extract R → Sketch shape(s) → Fit check → Spike unknowns → Iterate

Proactive Guidance During Shaping

After extracting initial R and Shape A, immediately show the fit check:

"Here's R x A — the fit between requirements and the proposed solution."

Then suggest based on what the fit check reveals:

Fit check showsSuggest
Flagged unknowns (warning markers)"A2 is flagged — want me to spike it?"
Requirements failing"R3 fails because [reason]. Should we revise the shape or adjust R3?"
All green"The shape looks solid. Ready to breadboard?"
Multiple viable shapes"Both A and B pass. Want to compare with show me A x R vs B x R?"

Shorthand Commands

Teach the user these as you go:

CommandWhat it does
show me R x AFit check: requirements against shape A
show me A x RRotated fit check: shape A against requirements
spike A2Investigate a specific part in depth
add R that...Add a new requirement
sketch shape BPropose an alternative approach

Phase complete when: A shape is selected (passes fit check, unknowns resolved).

"Shape [X] is selected and all requirements pass. Ready to breadboard?"

Phase 3: Breadboard

Detail the selected shape into concrete affordances with wiring.

  1. Invoke /breadboarding
  2. For existing projects: breadboard the mixture — existing + new affordances wired together
  3. For new projects: breadboard from the shape's parts
  4. Produce:

- Places table - UI Affordances table - Code Affordances table - Wiring diagram (Mermaid)

After completing:

"The breadboard shows [N] UI affordances, [N] code affordances across [N] places. Ready to slice into vertical increments?"

Phase 4: Slice

Cut the breadboard into demoable vertical slices.

  1. Invoke /breadboarding slicing procedure
  2. Identify V1 as the minimal demoable increment
  3. Layer additional capabilities as V2..V9
  4. For each slice, define:

- Which affordances it includes - What mechanism it demonstrates - A concrete demo statement

After completing: Present the slice summary and say:

"We have [N] slices. V1 demos '[demo statement]'. Want to start building V1?"

Phase 5: Build

Implement each slice, verify it works, then move to the next.

For each slice:

  1. Create an implementation plan (V[N]-plan.md)
  2. Build the slice
  3. Verify: run and test to confirm the demo works
  4. Update the Big Picture document with completion status

After each slice:

"V[N] is complete — [demo statement] is working. Ready for V[N+1]?"

File Management

FilePurposeLocation
shaping.mdGround truth for R, shapes, fit checks.claude/ or project root
spike-[topic].mdInvestigation findings.claude/ or project root
big-picture.mdHigh-level summary of entire feature.claude/ or project root
V[N]-plan.mdPer-slice implementation plans.claude/ or project root

Resuming a Session

When invoked with an existing shaping.md:

  1. Read the shaping doc to understand current state
  2. Invoke /shaping — it will detect the existing doc and show the fit check for the selected shape
  3. Identify what needs attention (unsolved requirements, pending slices)
  4. Suggest the next action:

- If shaping: "R4 is still undecided — want to discuss?" - If breadboarded: "Ready to slice?" - If sliced: "V3 is next — want to build it?" - If building: "V2 is complete. V3 demos '[statement]' — ready?"

平台分布

Codex

36.78%
按下载量换算61

Claude

31.47%
按下载量换算52

Cursor

17.9%
按下载量换算30

Gemini CLI

9.2%
按下载量换算15

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

需要联网

该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills