Token导航 LogoToken导航

Agent Skill · 前端设计

rubber-duck(橡皮鸭)

rubber-duck 用于处理 GitHub 仓库、Issue、Pull Request 和代码协作信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要围绕仓库状态、代码变更或协作事项进行整理时使用。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。

rubber-duck 是前端设计类 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,按安装量在前端设计类的 10,216 个 Skill 里排第 9,254。

总安装 222周安装 9GitHub Stars 公开资料未说明下载 70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:rubber-duck(橡皮鸭)
来源仓库:https://github.com/ouj/skills
仓库路径:skills/rubber-duck
安装命令:
npx skills add https://github.com/ouj/skills --skill rubber-duck
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/ouj/skills --skill rubber-duck

这个 Skill 是做什么的

前端设计类 Skill 关注页面结构、视觉规范和组件写法,输出通常是可运行的界面代码。

rubber-duck 用于处理 GitHub 仓库、Issue、Pull Request 和代码协作信息。适合在 Codex、Claude、Cursor、Gemini CLI 中围绕仓库状态或代码变更进行整理。通过 npx skills add 命令从指定仓库安装,需结合 README 确认具体用法。

安装前建议核实权限范围、维护状态及是否涉及联网或文件操作。适用宿主包括 Codex、Claude、Cursor、Gemini CLI,接入前应确认版本、权限和运行环境要求。

来源仓库内路径:skills/rubber-duck

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/ouj/skills --skill rubber-duck;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

rubber-duck 怎么安装?

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

rubber-duck 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。

rubber-duck 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。

rubber-duck 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。

rubber-duck 用的人多吗?

公开数据是累计安装 222 次,近一周安装 9 次,下载 70 次。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。

rubber-duck 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。
  • 需要处理前端设计类任务、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。

前端设计类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
frontend-design

前端设计 · Codex、Claude

8,823,360126,072只读MIT通过
vercel-composition-patterns

Vercel 组合模式 · Codex、Claude

3,858,81625,893需要联网MIT通过
vercel-react-native-skills

Vercel React native skills 前端 · Codex、Claude

2,468,66425,878只读MIT通过
adapt

adapt 搜索 · Codex、Claude

2,027,04023,125操作浏览器未明确通过
bolder

bolder 分析 · Codex、Claude

2,004,28823,130只读未明确通过
rubber-duck

橡皮鸭 · 本页

222—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

Rubber Duck

Explore an idea until it becomes a clean, minimal, well-structured design. Prefer simplicity, elegance, small change surfaces, and interface clarity over implementation detail. Aim for a strong solution direction, not a production-ready delivery plan.

Workflow

Follow these steps in order. Do not skip ahead unless the user explicitly narrows the scope or the codebase answers the open question.

  1. Explore project context
  • Inspect relevant files, docs, and recent commits before asking detailed questions.
  • Summarize the current system briefly: important components, relevant interfaces, existing constraints, and patterns worth preserving.
  • If the request is too broad, decompose it into smaller sub-problems and guide the user toward designing one slice at a time.
  1. Ask clarifying questions
  • Ask one question per message.
  • Always use lettered multiple-choice options.
  • Include a recommended option and a short reason.
  • Focus on purpose, constraints, success criteria, user impact, and interface impact.
  • If codebase exploration can answer the question, inspect the code instead of asking.

Use this format:

A. Recommended -... B.... C.......

  1. Propose approaches
  • Present 2-3 viable approaches.
  • For each approach, explain what changes, why it is attractive, what it complicates, and how it affects existing interfaces or workflows.
  • Also allow freeform input outside proposed approaches.
  • Lead with the recommended approach.
  • Prefer the smallest change that cleanly solves the problem.
  • Apply YAGNI aggressively. Remove features, cases, and moving parts that are not required.
  • Explore alternative branches when the user wants to compare directions before deciding.
  1. Align on the current system
  • Explain the relevant change surfaces and system boundaries before describing the new design.
  • Draw a "before" interaction diagram when the system has more than one component or boundary.
  • Use Mermaid when it improves clarity.
  1. Align on the new system
  • Show the "after" interaction diagram.
  • Call out interface changes explicitly: UI states, API contracts, props, events, routes, schema fields, config shape, CLI flags, or cross-service messages.
  • Prefer code-shaped examples for interface changes when that is clearer than prose, but mark them as illustrative only.
  • Keep illustrative snippets small: diffs, signatures, JSON, JSX, SQL, or type definitions are usually enough.
  1. Write the design doc
  • Write the design doc only after alignment on the design.
  • Include diagram and visualizations.
  • Keep it concise and structured.

Suggested sections:

  • Problem
  • Goals
  • Non-goals
  • Current state
  • Approaches considered
  • Recommended design
  • Before diagram
  • After diagram
  • Interface changes
  • Open questions and trade-offs
  1. Self-review the spec
  • Check for placeholders.
  • Check for contradictions between sections.
  • Check for ambiguous wording such as "simple", "fast", or "better" without measurable meaning.
  • Check that scope is minimal and matches the approved design.
  • Check that interface changes are called out explicitly.
  • Check that the spec stays focused on behavior, interfaces, and trade-offs rather than implementation detail or delivery planning.
  • Fix issues inline before presenting the final spec.

Design Boundaries

  • Break the system into units with one clear purpose.
  • Prefer boundaries with explicit interfaces and understandable dependencies.
  • Be able to explain what each unit does, how it is used, and what it depends on without diving into internals.
  • If a boundary is hard to explain cleanly, simplify the design before moving forward.

Working Style

  • Keep the conversation collaborative and natural, but disciplined.
  • Present the design in sections when the problem is complex and get explicit approval after each substantial section before moving on.
  • Prefer interface and user-visible behavior over low-level implementation detail.
  • Use code examples to explain contracts and interactions when that is clearer than prose.
  • Stay grounded in the existing codebase and patterns.
  • If the current codebase has local structural issues that materially affect the design, include small targeted improvements in the design.
  • Do not propose unrelated refactors.
  • Do not spend time on tests, rollout plans, metrics, monitoring, or operational readiness unless the user explicitly asks for them.
  • Keep the discussion centered on behavior, interfaces, boundaries, and trade-offs.
  • After the user approves the design, stop at the spec unless the user explicitly asks to move into implementation.

平台分布

Codex

38.96%
按下载量换算27

Claude

28.65%
按下载量换算20

Cursor

20.7%
按下载量换算14

Gemini CLI

9.63%
按下载量换算7

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