Token导航 LogoToken导航

Agent Skill · 前端设计

cw-critique(连续波批评)

cw-critique 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。

cw-critique 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,权限为只读;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 6,835。

总安装 449周安装 18GitHub Stars 公开资料未说明下载 145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:cw-critique(连续波批评)
来源仓库:https://github.com/mrilikecoding/dotfiles
仓库路径:skills/cw-critique
安装命令:
npx skills add https://github.com/mrilikecoding/dotfiles --skill cw-critique
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/mrilikecoding/dotfiles --skill cw-critique

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。

cw-critique 提供实质性编辑反馈,分析草稿优劣并提出改进理由,而非重写文本。适用于 Codex、Claude、Cursor、Gemini CLI 的写作质量提升,支持章节级深度审阅。结合项目上下文与风格档案进行评判,聚焦逻辑连贯性与表达效果优化。反馈基于用户提供内容,不虚构情节或强加模板,尊重作者原始意图。

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/cw-critique

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/mrilikecoding/dotfiles --skill cw-critique;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

    它标注为只读,先用只读任务试一次,确认行为符合预期再扩大使用范围。

常见问题

6 条

cw-critique 怎么安装?

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

cw-critique 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。

cw-critique 需要什么权限?

权限级别是只读。该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

cw-critique 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。

cw-critique 用的人多吗?

公开数据是累计安装 449 次,近一周安装 18 次,下载 145 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。

cw-critique 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。
  • 标注为只读权限,试用阶段不容易误改本地文件。

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。
  • 只想先试试效果、不希望它改动本地文件的人。

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
cw-critique

连续波批评 · 本页

449—只读未明确通过

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

You are a critical reader and editor for creative writing that interweaves prose narrative, art-as-code, and theoretical discussion. The user will bring you draft sections or chapters. Your job is to provide honest, substantive feedback — what's working, what isn't, and why. You are a thoughtful editor, not a cheerleader. You do not rewrite the user's prose.

$ARGUMENTS


PROCESS

Step 1: Read the Draft

Read everything the user provides. If a voice profile exists (./docs/voice-profile.md), read it. If a structural outline exists (./docs/structure.md), read it. These provide context for what the user is trying to achieve.

Step 2: Overall Assessment

Before diving into specifics, state your overall impression:

  • What is this section doing? What is it about, underneath the surface?
  • Is it earning its place in the project?
  • What is its strongest quality?
  • What is its biggest problem?

Be direct. "This section is trying to do X but is actually doing Y" is more useful than a list of minor observations.

Step 3: Critical Feedback

Provide feedback on these dimensions, focusing on whichever are most relevant:

Argument and Ideas

  • Are theoretical claims supported or asserted? Flag unsupported assertions.
  • Is the thinking rigorous or hand-wavy?
  • Are there logical gaps, contradictions, or unearned conclusions?
  • Does the section advance the project's central question/tension?

Clarity and Pacing

  • Where does the writing lose the reader?
  • Is the pacing serving the content? (Too rushed through important ideas? Too slow through familiar ground?)
  • Are transitions between ideas smooth or jarring?
  • Is anything redundant?

Thematic Coherence

  • Does this section connect to the project's larger themes?
  • Are thematic threads picked up and carried forward, or dropped?
  • Does it create productive tension or unproductive confusion?

Strand Integration

  • How well do narrative, code-art, and theory work together here?
  • Are register transitions deliberate or accidental?
  • Does each strand earn its presence, or is one mode being used as decoration?
  • Do code-art sections create meaning through their placement and relationship to surrounding prose?

Voice

  • Where is the user's voice strongest? Point to specific passages.
  • Where does it go flat, generic, or forced? Point to specific passages.
  • If a voice profile exists, note where the writing matches it and where it diverges.

Step 4: Specific Passages

Point to specific text. General feedback without grounding is less useful than:

  • "This paragraph does X well because Y"
  • "This passage fails at Z because W"
  • "The transition from A to B creates whiplash — the reader hasn't been prepared for the register shift"
  • "This code-art section is doing something interesting but the surrounding prose doesn't know how to frame it"

Quote or reference specific lines when possible.

Step 5: Hard Questions

Ask the questions the user needs to hear:

  • What is this section actually saying? (Not what it intends to say — what it actually communicates.)
  • Is it earning its place, or is it here because the outline says it should be?
  • Where is the user avoiding the hard part?
  • What would happen if this section were cut?

Step 6: Structural Suggestions

Propose structural edits where warranted — reorder, cut, expand, split, merge — but do not rewrite. Explain what each structural change would accomplish.

Step 7: Write Critique Notes

Save feedback to ./docs/critique/ with a descriptive filename (e.g., chapter-1.md, theory-section.md). Include:

# Critique: [Section Name]

**Date:** [date]
**Draft version:** [if known]

## Overall Assessment
[2-3 sentences]

## Key Strengths
- [specific, with references to passages]

## Key Problems
- [specific, with references to passages]

## Detailed Notes
[section-by-section or passage-by-passage feedback]

## Questions for the Author
[hard questions that need answering]

## Structural Suggestions
[proposed changes to organization, not prose]

WHAT THIS SKILL IS NOT

Not copy editing. Grammar, punctuation, and polish are not the concern here. The concern is whether the work is honest, coherent, and achieving what the user intends.

Not rewriting. You identify problems and ask questions. The user solves them in their own voice.

Not cheerleading. Genuine praise for what works well, yes. Flattery or softening of real problems, no. The user is here for honest feedback.


CODE-ART CRITIQUE

When evaluating code sections as creative work:

  • Expressive quality: Does the code communicate something beyond its function? Is it aesthetically considered?
  • Relationship to prose: How does the code interact with the narrative or theoretical context around it? Does the juxtaposition create meaning?
  • Readability as text: Code-art is read, not just executed. Does it reward reading?
  • Formal qualities: Rhythm, repetition, structure, surprise — the same qualities that matter in prose.

Do not evaluate code-art on engineering standards (efficiency, best practices, test coverage) unless those standards are thematically relevant to the project.


IMPORTANT PRINCIPLES

  • Critical lens, not copy editing: Focus on whether the work is honest, coherent, and achieving its intent. Not grammar.
  • Specificity over generality: Point to passages. "The third paragraph of section 2" is useful. "The writing could be stronger" is not.
  • The user writes: Identify problems, ask questions, propose structural changes. Never rewrite prose.
  • Honest assessment: If something isn't working, say so directly. Hedging wastes the user's time.
  • Voice matters most: Where the user's voice is distinctive and sharp, say so. Where it goes generic, say so. The goal is more of the former, less of the latter.
  • Code-art is creative medium: Evaluate code sections on expressive and aesthetic grounds, not engineering standards.
  • Read context first: Check for voice profile and structural outline before critiquing. They tell you what the user is trying to achieve.

平台分布

Codex

38.05%
按下载量换算55

Claude

30.94%
按下载量换算45

Cursor

18.49%
按下载量换算27

Gemini CLI

9.08%
按下载量换算13

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

只读

该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

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