Token导航 LogoToken导航

Agent Skill · 前端设计

gsp-style(普惠制风格)

gsp-style 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。

gsp-style 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 8,262。

总安装 303周安装 13GitHub Stars 29下载 106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:gsp-style(普惠制风格)
来源仓库:https://github.com/jubscodes/get-shit-pretty
仓库路径:skills/gsp-style
安装命令:
npx skills add https://github.com/jubscodes/get-shit-pretty --skill gsp-style
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/jubscodes/get-shit-pretty --skill gsp-style

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。

gsp-style 用于处理 GitHub 仓库、Issue、Pull Request 和代码协作信息,适合整理项目状态和变更事项。适用于围绕仓库状态、代码变更或协作事项进行信息组织和分析。通过 npx skills add 命令从 GitHub 仓库安装并使用。

安装前需确认权限范围和维护状态,注意可能触发联网或文件操作。gsp-style 属于前端设计类 Skill,可作为该场景下的辅助能力补充。

来源仓库内路径:skills/gsp-style

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/jubscodes/get-shit-pretty --skill gsp-style;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

gsp-style 怎么安装?

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

gsp-style 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。

gsp-style 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。

gsp-style 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。

gsp-style 用的人多吗?

公开数据是累计安装 303 次,近一周安装 13 次,下载 106 次,GitHub 29 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。

gsp-style 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
gsp-style

普惠制风格 · 本页

30329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

This is a standalone composable skill. It works two ways:

  1. Standalone — user runs /gsp-style cyberpunk directly, gets visual preview + tokens
  2. As a building block — agents invoke this skill during workflows, getting tokens only

Input: Style name(s), optional flags (--list, --preview) Output: {preset-name}.yml + STYLE.md + INDEX.md in the target system directory Agent: None — token expansion from YAML presets is handled inline

<execution_context> @${CLAUDE_SKILL_DIR}/styles/INDEX.yml @${CLAUDE_SKILL_DIR}/../gsp-brand-guidelines/design-tokens.md @${CLAUDE_SKILL_DIR}/chunk-format.md @${CLAUDE_SKILL_DIR}/../../templates/phases/patterns.md </execution_context>

Read the user's input to determine the mode:

InputMode
/gsp-style --listList all available presets
/gsp-style --preview cyberpunkShow tokens without writing files
/gsp-style cyberpunkApply single style
/gsp-style cyberpunk + neubrutalismMix styles (check compatibility)
/gsp-style (no args)Interactive — browse and pick

Step 1: List mode (--list)

Read styles/INDEX.yml and display all presets grouped by category. Format each as {name} {one-line description} with ─── Category ──── separators and footer showing usage. Example:

  ─── Minimal ────────────────────────
    swiss-minimalist     Helvetica and whitespace — let the content breathe
    flat-design          Solid colors, zero shadows — bold color blocks as structure

Stop here. Do not write any files.

Step 2: Interactive mode (no args)

If no style name provided, show the list (Step 1) then use AskUserQuestion with options grouped by category. Present 6-8 curated options spanning different categories, plus a "Show all" option:

  • neubrutalism — "thick borders, hard shadows, unapologetically flat"
  • modern-dark — "Linear/Vercel aesthetic — ambient blobs, mouse spotlights"
  • professional — "clean, trustworthy, gets out of the way"
  • cyberpunk — "neon-soaked interfaces from a dystopian future"
  • botanical — "deep greens, paper grain, nature-inspired"
  • claymorphism — "vinyl toy aesthetic — 4-layer shadows, squish"
  • liquid-glass — "Apple's 2025 — refractive, fluid, alive"
  • Show all 34 — "browse the full catalog"

Continue with the selected style.

Step 3: Load preset(s)

Read the YAML preset file(s) from styles/{name}.yml.

If the style name doesn't match any preset, suggest the closest match based on tags in INDEX.yml. If still no match, tell the user and show the list.

Mixing styles

If multiple styles are specified (e.g., cyberpunk + bento-grid):

  1. Check INDEX.yml clash_pairs — if the combination appears, warn the user and stop: "These styles clash — {reason}. Pick one or try a compatible combination."
  2. If compatible, merge token objects with last-wins precedence. The second style's values override the first where they overlap.

Step 4: Resolve output path

Determine where to write the system output:

Within a brand

If a brand context exists (.design/branding/{brand}/):

  • Write to {BRAND_PATH}/patterns/
  • This replaces the patterns phase of the branding diamond

Within a project (quick mode)

If invoked from a project context (.design/projects/{project}/):

  • Check if a brand.ref exists pointing to a brand with a completed system
  • If no brand system exists, write to .design/branding/_style-{preset-name}/patterns/
  • Create a minimal brand directory with just the system output
  • Update the project's brand.ref to point to this auto-generated brand

Standalone (no.design/ context)

  • Write to .design/branding/_style-{preset-name}/patterns/
  • Create minimal brand directory structure

Step 5: Preview mode (--preview)

If --preview, display expanded tokens grouped by section (Color, Typography, Shape, Elevation, Motion) as key-value pairs. Footer: usage hint to apply. Stop here — do not write any files.

Step 6: Copy preset as brand style

Copy the preset .yml to the output path as the brand's style source:

  • If within a brand: {OUTPUT_PATH}/{preset-name}.yml
  • If standalone: {OUTPUT_PATH}/_style-{preset-name}.yml

If a .yml already exists at the output path, use AskUserQuestion: "A style preset already exists — overwrite?" with options Overwrite and Cancel. If cancelled, skip and proceed.

The .yml IS the token source of truth — no separate tokens.json needed. Token names in .yml map 1:1 to shadcn/ui CSS variable names. Run node bin/theme-css.js {preset-name}.yml to generate a ready-to-paste :root/.dark CSS block.

Step 7: Write STYLE.md

Read the style template from ${CLAUDE_SKILL_DIR}/../../templates/phases/style.md.

Read BOTH source files:

  • styles/{name}.yml — structured data (tokens, intensity, patterns, constraints, effects)
  • styles/{name}.md — prose companion (design philosophy, CSS code hints, component styling, textures).

If the .md companion doesn't exist, render STYLE.md from .yml data only (thinner but functional).

Render into the template sections:

  • Intensity — from .yml intensity: block
  • Philosophy — extract from .md companion's Design Philosophy section. Condense to 2-4 sentences capturing the emotional DNA and cultural references.
  • Patterns — from .yml patterns: block, rendered as tables per component
  • Constraints — from .yml constraints: block, rendered as never/always bullet lists
  • Effects — from .yml effects: block, rendered as interaction vocabulary + state tables
  • Bold Bets — extract from .md companion's "Non-Genericness" or "Bold Factor" section. Pick 3-5 most distinctive techniques with implementation specifics.
  • Implementation — extract from .md companion's component stylings and CSS code:

- Component Code Hints — Tailwind/CSS patterns beyond the Patterns tables - Textures & Surfaces — CSS for noise, halftone, grain (skip if style has none) - Typography Treatments — text-stroke, tracking overrides (skip if standard) - Animation Recipes — keyframes, transitions (skip if effects vocabulary is sufficient)

Write to {OUTPUT_PATH}/STYLE.md.

Step 8: Write INDEX.md

Write {OUTPUT_PATH}/INDEX.md — header with phase/style/date, applied style name + description, file table ({preset-name}.yml, STYLE.md).

Foundation chunks (color-system.md, typography.md, etc.) are NOT written in the quick path — the .yml has the token values, STYLE.md has the composition rules. Foundation chunks are only produced by the full branding diamond where the brand-engineer adds deeper analysis (WCAG ratios, font loading, semantic rationale).

Step 9: Update state

If a brand STATE.md exists at the brand path:

  • Set patterns phase status to complete
  • Record style preset name and completion date
  • Set Prettiness Level to 60% (foundations only, no components)

If a project config.json exists:

  • Add "style_preset": "{name}" to preferences

Step 10: Completion output

Show: header (/gsp-style — {name} applied), file tree ({name}.yml + STYLE.md + INDEX.md). Then AskUserQuestion: Start a project → /gsp-project-brief, Build components → /gsp-brand-guidelines, Preview tokens, Try a different style → restart Step 2.

平台分布

Codex

38.24%
按下载量换算41

Claude

27.17%
按下载量换算29

Cursor

18.37%
按下载量换算19

Gemini CLI

9.69%
按下载量换算10

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