web-quality-skills MCP — 사용 사례, 설치 & 라이브 데모

왜 쓰나요

핵심 기능

6 skills covering the Lighthouse + Core Web Vitals matrix
Explicit targets: LCP ≤ 2.5s, INP ≤ 200ms, CLS ≤ 0.1
WCAG 2.2 accessibility auditing
Framework-agnostic recommendations
Auto-activation on matching requests

라이브 데모

실제 사용 모습

web-quality-skill.replay ▶ 준비됨

0/0

설치

클라이언트 선택

~/Library/Application Support/Claude/claude_desktop_config.json · Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "web-quality-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/addyosmani/web-quality-skills",
        "~/.claude/skills/web-quality-skills"
      ],
      "_inferred": true
    }
  }
}

Claude Desktop → Settings → Developer → Edit Config 열기. 저장 후 앱 재시작.

~/.cursor/mcp.json · .cursor/mcp.json

{
  "mcpServers": {
    "web-quality-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/addyosmani/web-quality-skills",
        "~/.claude/skills/web-quality-skills"
      ],
      "_inferred": true
    }
  }
}

Cursor는 Claude Desktop과 동일한 mcpServers 스키마 사용. 프로젝트 설정이 전역보다 우선.

VS Code → Cline → MCP Servers → Edit

{
  "mcpServers": {
    "web-quality-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/addyosmani/web-quality-skills",
        "~/.claude/skills/web-quality-skills"
      ],
      "_inferred": true
    }
  }
}

Cline 사이드바의 MCP Servers 아이콘 클릭 후 "Edit Configuration" 선택.

~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "web-quality-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/addyosmani/web-quality-skills",
        "~/.claude/skills/web-quality-skills"
      ],
      "_inferred": true
    }
  }
}

Claude Desktop과 같은 형식. Windsurf 재시작 후 적용.

~/.continue/config.json

{
  "mcpServers": [
    {
      "name": "web-quality-skill",
      "command": "git",
      "args": [
        "clone",
        "https://github.com/addyosmani/web-quality-skills",
        "~/.claude/skills/web-quality-skills"
      ]
    }
  ]
}

Continue는 맵이 아닌 서버 오브젝트 배열 사용.

~/.config/zed/settings.json

{
  "context_servers": {
    "web-quality-skill": {
      "command": {
        "path": "git",
        "args": [
          "clone",
          "https://github.com/addyosmani/web-quality-skills",
          "~/.claude/skills/web-quality-skills"
        ]
      }
    }
  }
}

context_servers에 추가. 저장 시 Zed가 핫 리로드.

claude mcp add web-quality-skill -- git clone https://github.com/addyosmani/web-quality-skills ~/.claude/skills/web-quality-skills

한 줄 명령. claude mcp list로 확인, claude mcp remove로 제거.

사용 사례

실전 활용법: web-quality-skills

Fix a slow LCP on a product landing page

👤 Frontend engineers trying to get a green CWV ⏱ ~40 min intermediate

언제 쓸까: CrUX shows LCP at 3.8s and you don't know where to start.

사전 조건

Target URL — Live or staging URL — not just local

흐름

Run the core-web-vitals skill

Use the core-web-vitals skill on https://site.com. Focus on LCP.✓ 복사됨

→ Diagnosis of the LCP element (hero image, web font, server response) with specific fix
Apply the fix

Apply the recommended priority hints and image format change.✓ 복사됨

→ Concrete HTML changes
Re-measure

How do I verify the improvement?✓ 복사됨

→ Guided re-measure via Lighthouse or web-vitals JS

결과: LCP under the 2.5s target with a documented fix path.

함정

Fixing lab LCP but not field LCP (CrUX) — Always verify in CrUX, not just local Lighthouse

Run a WCAG 2.2 accessibility audit on a checkout flow

👤 Engineers preparing for an accessibility review ⏱ ~45 min intermediate

언제 쓸까: Legal is asking about WCAG compliance for the checkout.

흐름

Run the accessibility skill

Use the accessibility skill on our checkout flow. What WCAG 2.2 issues exist?✓ 복사됨

→ Structured findings per WCAG success criterion with severity
Prioritize fixes

Group by fix-effort vs user impact.✓ 복사됨

→ Ordered backlog

결과: A real accessibility backlog, not a checklist you can't act on.

함정

Automated checks only catch ~30% of issues — Complement with keyboard-only and screen-reader manual testing

함께 쓰기: playwright-skill

Run the full web-quality-audit before a marketing campaign

👤 Teams about to drive paid traffic ⏱ ~30 min intermediate

언제 쓸까: Big campaign starts Monday and you want no nasty surprises.

흐름

Run the combined audit

Use web-quality-audit on our campaign landing page.✓ 복사됨

→ Report spanning performance, CWV, accessibility, SEO, best-practices
Triage for launch

What must be fixed before launch vs can wait?✓ 복사됨

→ Go/no-go triage

결과: A launch-ready page and a week-2 backlog.

함정

Treating every finding as blocker — The skill prioritizes — trust the must-fix list

조합

다른 MCP와 조합해 10배 효율

web-quality-skill + chrome-devtools

Use DevTools MCP to grab live perf traces while the skill interprets them

Capture a perf trace via chrome-devtools, then have the performance skill analyze it.✓ 복사됨

web-quality-skill + playwright-skill

Playwright runs the site through real user flows, web-quality skills score each step

Use playwright to walk through checkout, capture metrics at each step, then accessibility-audit each screen.✓ 복사됨

web-quality-skill + claude-seo-skill

Technical SEO from claude-seo + CWV from this skill

Run /seo audit for link/schema issues, then web-quality-audit for CWV.✓ 복사됨

도구

이 MCP가 노출하는 것

도구	입력	언제 호출
web-quality-audit	URL	Pre-launch or quarterly
performance	URL	Slow page
core-web-vitals	URL	Failing CWV
accessibility	URL / component	A11y review
seo	URL	SEO hygiene check
best-practices	URL	Hardening review

비용 및 제한

운영 비용

API 쿼터: None
호출당 토큰: 2-5k per skill loaded
금액: Free — skill is local
팁: Audit one URL at a time; batch audits balloon context.

보안

권한, 시크릿, 파급범위

자격 증명 저장: No credentials — skill is prompts

데이터 외부 송신: None from the skill itself

문제 해결

자주 발생하는 오류와 해결

Skill gives generic advice

Feed in a real Lighthouse JSON or CrUX record for specific findings.

Recommendations don't apply to your framework

Name the framework ('Next.js 15 app router') so the skill tailors suggestions.

대안

web-quality-skills 다른 것과 비교

대안	언제 쓰나	단점/장점
Lighthouse CI directly	You want automated CI gates, not interactive guidance	Scalable for CI; less explanatory
claude-seo-skill	You need SEO-first with broader audit surface	Broader scope but less CWV depth

web-quality-skills