- Home
- Skills
- Frontend & Web Apps
- Web Performance HTML Check
More screenshots
Works with the AI Agents you already use
Web Performance HTML Check
Find what slows LCP and makes pages jump in HTML, templates, React, Vue, Svelte, Astro and CSS: lazy hero images, broken preloads, sync head scripts, fonts without font-display, unsized images, heavy images and embeds.
$5
Web Performance HTML Check
Example session with this skill installed
Check this project with web-performance-html-check.
- Read your context and instructions
- Compiled the web performance html
- Generated the UI component
$ python3 scripts/check_webperf.py examples/risky
BLOCK shop/app/page.tsx:13 P01 PRELOADED_IMAGE_LAZY_LOADED
loading="lazy"
Why: This image is preloaded (a link rel="preload" as="image" for the same file in this page, or the priority or preload prop of next/image, Astro's Image or Angular's NgOptimizedImage) and also has loading="lazy", and no other copy of that image in the file loads eagerly. The two contradict each other: the preload says the page needs this image first, usually because it is the Largest Contentful Paint (LCP) image, while lazy loading makes the browser wait until layout is done before it loads and shows it. web.dev: never lazy-load your LCP image.
Safer: Remove loading="lazy" from this image (or set loading="eager") and keep the preload or fetchpriority="high". Keep lazy loading for the images further down the page.
BLOCK site/index.html:10 P03 FONT_PRELOAD_WITHOUT_CROSSORIGIN
BLOCK site/index.html:11 P02 PRELOAD_WITHOUT_VALID_AS
BLOCK site/index.html:23 P01 PRELOADED_IMAGE_LAZY_LOADED
BLOCK site/js/ads.js:3 P04 DOCUMENT_WRITE_EXTERNAL_SCRIPT
REVIEW shop/app/page.tsx:15 P11 LAZY_IMAGE_LIKELY_ABOVE_THE_FOLD
REVIEW site/css/site.css:1 P14 FONT_WITHOUT_FONT_DISPLAY
REVIEW site/css/site.css:1 P15 CSS_IMPORT_CHAIN
REVIEW site/css/site.css:2 P15 CSS_IMPORT_CHAIN
REVIEW site/css/site.css:4 P14 FONT_WITHOUT_FONT_DISPLAY
REVIEW site/css/site.css:19 P16 LARGE_INLINE_DATA_URI
REVIEW site/index.html:9 P14 FONT_WITHOUT_FONT_DISPLAY
REVIEW site/index.html:14 P13 RENDER_BLOCKING_SCRIPT
REVIEW site/index.html:15 P13 RENDER_BLOCKING_SCRIPT
REVIEW site/index.html:28 P12 IMAGE_WITHOUT_DIMENSIONS
REVIEW site/index.html:29 P19 SRCSET_WITHOUT_SIZES
REVIEW site/index.html:31 P18 LARGE_IMAGE_FILE
REVIEW site/index.html:35 P17 IFRAME_NOT_LAZY
(Each finding also has Why and Safer lines, as in the first one.)
Verdict: BLOCK. Checked 6 files, 6 with web page or CSS content.
This is a static check of file text. NOT FLAGGED is not a guarantee.
page.tsx
TSX · React component
Example file from a real run - the skill writes it into your workspace.
Connects securely to your tools. The creator never sees your data.
About this skill
A hero image with loading="lazy" next to a preload for the same file, a font preload without crossorigin that downloads the font twice, a link rel="preload" with no as value that the browser ignores: each one quietly costs your page its Largest Contentful Paint. So do images without width and height that make the text jump, a chat widget loaded synchronously in the head, and a 600 KB illustration nobody resized. This skill finds those and the other loading mistakes listed below before the page ships, and tells you how to fix each one.
Your agent runs a scanner on your HTML pages and templates (Liquid, Nunjucks, Handlebars, EJS, ERB, Twig), Vue, Svelte and Astro components, JSX/TSX, JavaScript that writes markup, and CSS. It reads tags the way frameworks write them (JSX expressions and spreads, Vue bound attributes, template blocks), knows that next/image, Astro's Image and Angular's NgOptimizedImage lazy-load unless marked priority, skips comments, noscript fallbacks and values that templates fill in at run time, and leaves the standard Google Tag Manager, gtag, Google Fonts (with display=swap) and Cookiebot snippets alone. For image files your pages use, it reads only the file size. It never runs, builds or loads your site. It returns BLOCK, REVIEW, or NOT FLAGGED, with the file, line, reason, and a safer approach, and each rule links to the web.dev, HTML standard or Chrome documentation behind it. Secrets are never printed.
What it checks
BLOCK
- P01: an image that is preloaded (link rel="preload", or priority/preload in next/image, Astro or Angular) and also loading="lazy", with no eager copy of it on the page
- P02: link rel="preload" without a valid as value (the browser fetches nothing)
- P03: font or fetch preloads without crossorigin (downloaded twice)
- P04: document.write on the page itself of a script from another host without async or defer
REVIEW
- P11: lazy images that look like the first screen (fetchpriority="high", first image in the body, or named hero, banner and similar)
- P12: img elements without width and height (sizes set in the page's own or linked CSS count)
- P13: scripts in the head without async, defer or type="module"
- P14: @font-face without font-display, and Google Fonts URLs without display=
- P15: CSS @import chains the browser has to follow one by one
- P16: data: URIs longer than 10,000 characters inlined in HTML, CSS or JavaScript
- P17: iframes (videos, maps, widgets) without a loading attribute
- P18: image files over 500 KB used by a page, component or stylesheet
- P19: srcset with width descriptors and no sizes
What's included
- The scanner (Python 3.9+, standard library only)
- A rules guide with the reason, safer approach and official source link for each rule
- A risky and a safe example project (a static site and a Next.js page), and the real report the scanner produces on the risky one
Limits It reads each file on its own and does not know screen sizes, so it cannot tell which image is really the LCP element: the first-screen rule is a hint to confirm. CSS the page does not link or import itself (a layout's global stylesheet, CSS modules, CSS-in-JS), preloads in a layout file or HTTP headers, and markup built at run time are not seen. It does not measure INP, server response time or the cost of third-party scripts; confirm the result with Lighthouse or PageSpeed Insights. NOT FLAGGED is not a guarantee.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Frontend & Web Apps.
Static check of page, component and CSS text plus the size of referenced image files. The real LCP element, network conditions and markup built in the browser are not measured. NOT FLAGGED is not a guarantee.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 4 days ago
Permissions
Runs one local Python script in the terminal. It reads your project's files, writes nothing, and makes no network calls. The example project includes a 600 KB SVG illustration (plain text) so the large-image rule has a file to find, and fake URLs and IDs. Nothing in it is run.