Token导航 LogoToken导航TokenDH.com
前端设计只读github未标认证来源可访问clear审计通过

micro-polish微抛光

Agent Skill

用于辅助前端页面、组件、样式和交互逻辑的开发与维护。它适合让 Agent 生成或审查 React、Next.js、Vue、Tailwind、CSS 等相关代码,整理组件结构,或定位布局和性能问题。使用时需要结合项目现有设计系统、路由和构建方式,避免只生成孤立片段;涉及页面改动时,应配合本地预览和构建检查确认视觉效果。

总安装

198

周安装

8

GitHub Stars

1

下载量

62
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

3

许可证

MIT

最后核验

2026-05-01

来源状态

来源可访问

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:micro-polish(微抛光)
来源仓库:https://github.com/justingoertz/micro-polish-skill
仓库路径:skills/micro-polish
安装命令:
npx skills add https://github.com/justingoertz/micro-polish-skill --skill micro-polish
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。不同来源提供的安装方式可能略有差异;本站展示可直接复制的安装命令,安装前请核对来源页面。

skills.shnpx skills
npx skills add https://github.com/justingoertz/micro-polish-skill --skill micro-polish

简介

辅助前端页面、组件、样式和交互逻辑的开发与维护。

  • 适合生成或审查 React、Next.js、Vue、Tailwind、CSS 等相关代码。
  • 使用时需结合项目设计系统、路由和构建方式,避免孤立片段。
  • 涉及页面改动时应配合本地预览和构建检查确认视觉效果。
  • micro-polish 属于前端设计类 Skill,可作为该场景下的辅助能力补充。

SKILL.md

Micro Polish

Apply obsessive attention to details others won't notice. The accumulated micro choices in typography, spacing, timing, and rhythm compound into a feeling of quality that registers subconsciously before it registers consciously.

How to use

  • /micro-polish Apply micro-refinement sensibility to any work in this conversation.
  • /micro-polish <file> Review the file for micro-polish opportunities and output:

- observation (quote the exact line/snippet) - why it matters (the felt difference) - a concrete refinement

Philosophy

Quality emerges from hundreds of micro-decisions that individually seem insignificant but collectively create work that feels different—even when observers can't articulate why. This skill is about developing and trusting intuition for those choices.

Understand the context

Before polishing, read the work and understand what it is and who it's for. Infer from the code or UI itself—file names, dependencies, existing patterns, visual language, and domain signals all reveal intent.

Different contexts demand different choices:

  • A healthcare app for older users needs generous contrast, larger touch targets, and clarity over cleverness. "Subtle" would be a failure mode here.
  • A luxury fashion site might prize restraint—quiet typography, refined spacing, atmosphere over accessibility. "Bold" might feel cheap.
  • A developer CLI wants zero decoration. Polish means removing friction, not adding finesse.
  • A children's game can be exuberant in ways that would feel juvenile elsewhere.

The same "refinement" can be right in one context and wrong in another. Maximum contrast is polish for an accessibility-first app. It's anti-polish for a meditation app. There are no universal rules—only appropriate choices for specific contexts.

When context is genuinely ambiguous and the right direction is unclear, ask. Otherwise, infer and proceed.

Process

Read the work slowly

Before changing anything, read the code or view the UI at half speed. Notice what creates friction. Notice what flows. Trust initial reactions before rationalizing them away.

Notice what matters here

Given what this work is and who it's for, what does "polish" mean in this specific context? What would make this feel more like itself—more aligned with its purpose? Let that understanding guide which details deserve attention.

What to output

After polishing, provide:

  1. Context understood — What you inferred about the work and its users. What "polish" means here.
  2. Changes made — Refinements applied, with the specific observation that prompted each.
  3. Rationale — Why each choice fits *this* context. Not generic craft principles—the specific reasoning for this work.
  4. Left alone — Details you noticed but chose not to change, and why they're right as-is for this context.

Trust your eye

Some refinements can't be justified by rules. A 1px shift that makes an icon "sit right." A 50ms timing adjustment that makes motion "feel better." A color that's "not quite" the computed value but reads correctly.

These intuitions are valid. They come from accumulated exposure to quality work. Over time, you'll develop your own sense of what matters—and the confidence to override any guideline when your eye tells you otherwise.

Document your choices anyway—not to justify, but to build a vocabulary for what you're sensing.

The standard

Polish until you can look at every part of the work and feel settled. Not excited, not proud—settled. The absence of small irritations. The quiet confidence that comes from knowing nothing was left approximate when it could have been precise.

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

04

需要参考平台分布和安装热度时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

补充不同宿主或平台的使用分布数据

能力 5

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Claude Code

45.28%
按下载量换算28

OpenCode

32.03%
按下载量换算20

windsurf

14.49%
按下载量换算9

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。

来源信息

继续浏览同类 Skills