Token导航 LogoToken导航

Agent Skill · 研究检索

aesthetic-system(审美系统)

aesthetic-system 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。

aesthetic-system 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,权限为操作浏览器;许可证信息还不明确,第三方扫描有提醒项,按安装量在研究检索类的 73,675 个 Skill 里排第 55,093。

总安装 490周安装 20GitHub Stars 8下载 157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:aesthetic-system(审美系统)
来源仓库:https://github.com/phrazzld/claude-config
仓库路径:skills/aesthetic-system
安装命令:
npx skills add https://github.com/phrazzld/claude-config --skill aesthetic-system
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/phrazzld/claude-config --skill aesthetic-system

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。

aesthetic-system 提供前端开发前的强制性设计与实现指导,结合 Gemini 网络调研与 Kimi 模型完成视觉落地。适用于需要避免 AI 同质化、追求真实优秀案例参考的 Web 或移动端组件开发任务。

必须先在 Gemini 中发起设计方向调研,再通过 MCP 将前端实现完全委托给 Kimi K2.5 模型执行。安装前应检查 GitHub 仓库权限、维护频率及是否涉及外部 API 调用或代码生成行为。注意该技能强制要求两阶段流程,不可跳过研究或绕过 Kimi 直接编码,否则可能导致风格偏差。

来源仓库内路径:skills/aesthetic-system

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/phrazzld/claude-config --skill aesthetic-syste…;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

    它需要操作浏览器,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

aesthetic-system 怎么安装?

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

aesthetic-system 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。

aesthetic-system 需要什么权限?

权限级别是操作浏览器。该 Skill 可能涉及浏览器控制能力,使用时可能读取或操作网页内容,需要在受控环境中确认权限边界。

aesthetic-system 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。

aesthetic-system 用的人多吗?

公开数据是累计安装 490 次,近一周安装 20 次,下载 157 次,GitHub 8 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。

aesthetic-system 安全吗?

第三方扫描有提醒项,说明存在需要人工确认的行为。安装前请打开源码看清它读了什么、联网到哪里。

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。

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。
  • 第三方扫描存在提醒项或未通过项,需要人工逐条确认。
  • 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
aesthetic-system

审美系统 · 本页

4908操作浏览器未明确有提醒

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

Aesthetic System

Phase 0: Research & Delegation (MANDATORY)

Step 1: Gemini Research

Before ANY frontend work, get design direction from Gemini:

gemini -p "I'm building [describe component/page]. Research:
1. Distinctive design approaches (avoid AI-slop aesthetics)
2. Real-world examples of excellent [component type]
3. Current typography/color/layout trends for this context
4. Unexpected alternatives to obvious solutions"

Why mandatory: Web grounding surfaces current trends, prevents convergence.

Step 2: Kimi Implementation (MANDATORY)

All frontend implementation MUST be delegated to Kimi K2.5 via MCP.

Kimi excels at frontend development and visual coding:

// Single component/page implementation
mcp__kimi__spawn_agent({
  prompt: `Implement [component] following these aesthetics:
- Typography: ${typography}
- Colors: ${palette}
- Layout: ${layoutDirection}
- Motion: ${motionGuidelines}
Constraints: Apply ui-skills rules. No Inter/Roboto. No purple gradients.
Output: React/Tailwind component in ${targetPath}`,
  thinking: true  // For complex implementations
})

// Parallel implementation (multiple components)
mcp__kimi__spawn_agents_parallel({
  agents: [
    { prompt: "Implement Hero section...", thinking: true },
    { prompt: "Implement Card component...", thinking: true },
    { prompt: "Implement Navigation...", thinking: true },
  ]
})

Workflow:

  1. Research direction → Gemini (web grounding, trends)
  2. Implement visuals → Kimi (Agent Swarm, parallel execution)
  3. Review quality → Claude (expert panel, quality gates)

Anti-pattern: Implementing frontend yourself instead of delegating to Kimi. Pattern: Research (Gemini) → Build (Kimi) → Review (Claude)

Design Thinking

Before coding, commit to a BOLD aesthetic direction:

  1. Purpose: What problem? Who uses it?
  2. Tone: Pick extreme (brutalist, luxury, playful, editorial, organic...)
  3. Differentiation: What's the ONE thing someone will remember?

CRITICAL: Bold maximalism and refined minimalism both work—the key is intentionality.

Core Principles

  • Typography: Distinctive fonts, not Inter/Roboto/Space Grotesk
  • Color: Dominant + accent, brand-tinted neutrals, no pure grays
  • Motion: One orchestrated moment > scattered micro-interactions
  • Layout: Asymmetry, overlap, diagonal flow, grid-breaking
  • Backgrounds: Gradients, noise, patterns—not solid colors

See: references/aesthetics-guidelines.md

Quality Bar: Stripe-Level Excellence

Reference these as quality exemplars:

  • Stripe: Obsessive typography, micro-interactions, developer delight
  • Linear: Keyboard-first, performance as design, smooth motion
  • Vercel: Minimalist clarity, dramatic contrast, confident hierarchy

The Gasp Test: Would users gasp at how stunning this is? If no, keep iterating.

Quality Checkpoints:

  • Typography that makes people stop and notice
  • Layout that surprises with intentionality
  • Motion that feels physically satisfying
  • Details that show obsessive care
  • Mobile experience that doesn't "suck"

Mobile Excellence (Separate Optimization)

Mobile is NOT just responsive—it requires separate design thinking.

Touch Libraries to Consider

  • @use-gesture/react: Touch, mouse, drag, pinch gestures unified
  • react-spring: Gesture-aware animations with physical springs
  • swiper: Touch slider with native feel
  • framer-motion: Gesture recognition + animation
  • @capacitor/haptics: Haptics for Capacitor/mobile apps

Mobile-Specific Quality Checks

  • Touch targets: 44x44px minimum (Apple HIG)
  • Swipe gestures: Natural, discoverable, satisfying
  • Haptic feedback: Confirm actions with tactile response
  • Pull-to-refresh: Physical bounce, not instant
  • Bottom navigation: Thumb-reachable actions
  • Momentum scrolling: Physics-based scroll

See: references/mobile-excellence.md

Anti-Convergence

YOU TEND TOWARD GENERIC OUTPUTS. Before implementing, ask:

  • "Have I used this font/color/layout recently?"
  • "What's a distinctive alternative?"
  • "Does this feel unique to THIS project?"

Vary every project: light/dark, font pairings, aesthetics, color approach.

See: references/anti-patterns.md

Advanced Techniques

When basic CSS isn't enough:

  • WebGL/Three.js: Living backgrounds, gradient meshes
  • GSAP/Framer Motion: Complex animation sequences
  • CSS Art: Pure CSS illustrations, clip-path magic
  • ASCII: Terminal/brutalist aesthetics
  • Iconify: 200k+ icons when Lucide isn't enough

See: references/advanced-techniques.md

Iterative Polish Pattern

For incremental refinement after initial build:

Quick passes (run /polish pass repeatedly):

  • Each pass makes ONE high-impact change
  • Compound improvements over 10+ passes
  • Separate desktop/mobile analysis each time
  • Stream Deck / macro friendly

Full polish (run /polish):

  • Multi-agent Design Council critique
  • 5 automated iterations with quality threshold
  • Use for foundation establishment or comprehensive audit

Overhaul mode (say "it sucks" or "overhaul"):

  • Aggressive transformation, no preservation instinct
  • Forces greenfield treatment regardless of maturity
  • 8 iterations max, dramatic improvements

References

  • references/implementation-constraints.md - MUST/NEVER rules for implementation (stack, components, animation, performance)
  • references/aesthetics-guidelines.md - Typography, color, motion, spatial composition
  • references/advanced-techniques.md - WebGL, animations, CSS art, icons, asset generation
  • references/anti-patterns.md - AI slop to avoid, convergence traps, variation mandate
  • references/dna-codes.md - DNA variation system for structural variety
  • references/banned-patterns.md - Explicit banned elements (hero badges, generic fonts, etc.)
  • references/browser-helpers.md - Browser automation for Coolors, Google Fonts, inspiration

平台分布

Codex

37.39%
按下载量换算59

Claude

30.92%
按下载量换算49

Cursor

20.14%
按下载量换算32

Gemini CLI

8.86%
按下载量换算14

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

可疑

权限和风险

操作浏览器

该 Skill 可能涉及浏览器控制能力,使用时可能读取或操作网页内容,需要在受控环境中确认权限边界。

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。来源安全扫描存在提醒或未通过项,安装前请自行复核。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