agentleFS
dembrandtdembrandt/dembrandt-skills/skills/performance-and-web-vitals/SKILL.md

performance-and-web-vitals

Skill194 starsChanged 4 months agoMITView on GitHub

Lighthouse audits and Core Web Vitals — LCP, CLS, INP. Use when optimising page load, layout shift, input delay, images, fonts or render-blocking resources.

  • Installs packages

The file

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 2 are distinctive, and 4 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 4

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Never use loading="lazy" on the LCP image — it delays the most important render.

    line 110
  2. Very specific:

    Always set width and height — prevents CLS

    line 190
  3. Very specific:

    Use loading="lazy" below the fold, never on LCP image

    line 191
  4. Distinctive:

    Use a CDN with automatic format conversion where possible

    line 194
  5. Very specific:

    Never block the main thread with synchronous <script> in <head>

    line 239
  6. Distinctive:

    Run Lighthouse in CI to catch regressions before deployment.

    line 245

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Performance and Web Vitals

Run a Lighthouse Audit

# CLI audit — outputs JSON and HTML report
npx lighthouse https://example.com --output html --output-path ./lighthouse-report.html

# Headless, useful in CI
npx lighthouse https://example.com --chrome-flags="--headless" --output json --output-path ./report.json

# Audit specific categories only
npx lighthouse https://example.com --only-categories=performance,accessibility,seo

Or open Chrome DevTools → Lighthouse tab → Analyse page load.

Target scores:

CategoryTarget
Performance≥ 90
Accessibility100
Best Practices≥ 95
SEO≥ 95

Core Web Vitals

LCP — Largest Contentful Paint

How fast does the main content appear?

Target: ≤ 2.5s

LCP measures when the largest visible element (hero image, heading, video poster) renders. It is the user's perception of "did the page load?"

Common causes and fixes:

CauseFix
Unoptimised hero imageUse WebP/AVIF, correct size, fetchpriority="high"
Image not preloaded<link rel="preload" as="image" href="hero.webp">
Render-blocking CSS/JSDefer non-critical JS, inline critical CSS
Slow server responseCDN, caching headers, edge delivery
Web font blocking renderfont-display: swap or optional
<!-- Preload LCP image -->
<link rel="preload" as="image" href="hero.webp" fetchpriority="high">

<!-- LCP image: no lazy loading -->
<img src="hero.webp" alt="..." fetchpriority="high" width="1200" height="600">

Never use loading="lazy" on the LCP image — it delays the most important render.


CLS — Cumulative Layout Shift

Does content jump around while loading?

Target: ≤ 0.1

CLS measures unexpected layout shifts — content moving after it has rendered. Caused by images without dimensions, late-loading ads, fonts swapping, or dynamic content injected above existing content.

Common causes and fixes:

CauseFix
Images without width/heightAlways set width and height on <img>
Web font swapUse font-display: optional or preload fonts
Dynamic content above foldReserve space with min-height on containers
Late-loading ads or embedsReserve fixed dimensions for ad slots
Animations that shift layoutAnimate transform only, never top/left/width/height
<!-- Always include dimensions -->
<img src="product.jpg" width="400" height="300" alt="...">
/* Reserve space for dynamic content */
.ad-slot { min-height: 250px; }

/* Animate transform, not layout properties */
.slide-in { transform: translateY(0); transition: transform 300ms; }

INP — Interaction to Next Paint

How quickly does the page respond to user input?

Target: ≤ 200ms

INP measures the delay between a user interaction (click, tap, keyboard) and the next visual update. High INP makes the UI feel sluggish or frozen.

Common causes and fixes:

CauseFix
Heavy JS on main threadBreak into smaller tasks, use requestIdleCallback
Large event handlersDebounce/throttle scroll and resize handlers
Synchronous DOM updatesBatch DOM writes with requestAnimationFrame
Third-party scripts blockingLoad third-party scripts with async or defer
React re-rendersMemoize with useMemo, useCallback, React.memo

Images

Images are the single biggest performance lever on most pages.

<!-- Modern formats with fallback -->
<picture>
  <source srcset="image.avif" type="image/avif">
  <source srcset="image.webp" type="image/webp">
  <img src="image.jpg" width="800" height="600" alt="..." loading="lazy">
</picture>

<!-- Responsive images -->
<img
  srcset="image-400.webp 400w, image-800.webp 800w, image-1200.webp 1200w"
  sizes="(max-width: 600px) 100vw, 50vw"
  src="image-800.webp"
  alt="..."
  width="800"
  height="600"
  loading="lazy"
>

Rules:

  • Always set width and height — prevents CLS
  • Use loading="lazy" below the fold, never on LCP image
  • Serve WebP or AVIF — typically 30–50% smaller than JPEG
  • Size images to their display size — do not serve 2000px image for a 400px slot
  • Use a CDN with automatic format conversion where possible

Fonts

Web fonts block rendering if not handled correctly.

<!-- Preconnect to font origin -->
<link rel="preconnect" href="https://fonts.googleapis.com">
<link rel="preconnect" href="https://fonts.gstatic.com" crossorigin>

<!-- Preload critical font file -->
<link rel="preload" href="/fonts/brand.woff2" as="font" type="font/woff2" crossorigin>
@font-face {
  font-family: 'Brand';
  src: url('/fonts/brand.woff2') format('woff2');
  font-display: swap;     /* show fallback immediately, swap when loaded */
  /* font-display: optional; — never swap, use fallback if not cached */
}
  • font-display: swap — good for headings, acceptable CLS
  • font-display: optional — zero CLS, font only used if cached (best for body text)
  • Subset fonts to the characters actually used — reduces file size by 60–80%

JavaScript

<!-- Defer non-critical scripts -->
<script src="analytics.js" defer></script>
<script src="chat-widget.js" async></script>

<!-- Module scripts are deferred by default -->
<script type="module" src="app.js"></script>
  • defer: executes after HTML parsed, in order — use for most scripts
  • async: executes as soon as downloaded, out of order — use for independent scripts (analytics)
  • Never block the main thread with synchronous <script> in <head>

Lighthouse CI (automated audits)

Run Lighthouse in CI to catch regressions before deployment.

# Install
npm install -g @lhci/cli

# Run
lhci autorun --upload.target=temporary-public-storage
# .lighthouserc.json
{
  "ci": {
    "assert": {
      "assertions": {
        "categories:performance": ["warn", { "minScore": 0.9 }],
        "categories:accessibility": ["error", { "minScore": 1.0 }],
        "categories:seo": ["warn", { "minScore": 0.95 }]
      }
    }
  }
}

Review Checklist

  • Lighthouse performance score ≥ 90
  • Lighthouse accessibility score = 100
  • LCP ≤ 2.5s — LCP image preloaded, no loading="lazy" on it
  • CLS ≤ 0.1 — all images have width and height, no layout-shifting animations
  • INP ≤ 200ms — no heavy synchronous JS on main thread
  • Images served as WebP or AVIF with correct dimensions
  • loading="lazy" on all below-fold images
  • Web fonts use font-display: swap or optional
  • Non-critical JS loaded with defer or async
  • Lighthouse CI configured to catch regressions in deployment pipeline

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.