- Home
- Skills
- Marketing & SEO
- SEO Audits And Fixers For Websites
Works with the AI Agents you already use
SEO Audits And Fixers For Websites
An SEO audit of your frontend code: indexability, metadata, canonicals, structured data, robots and sitemap, ranked P0 to P2, with each batch of fixes applied only after you approve it
$5
/moSubscription · cancel anytime
Secure checkout via Stripe
- Always the latest version
- Delivered live to your agent via MCP
- Cancel anytime, access ends at period end
SEO Audits And Fixers For Websites
Example session with this skill installed
Audit our Next.js 14 (app router) product page for SEO before we ship. What we have:
app/products/[slug]/page.tsx
export default async function Page({ params }) {
const p = await getProduct(params.slug);
return (
<main>
<div className="title">{p.name}</div>
<img src={p.image} />
<p>{p.description}</p>
<span>{p.price} EUR</span>
</main>
);
}
app/layout.tsx sets metadata = { title: 'My Shop' } for every route. There is no sitemap. public/robots.txt is:
User-agent: *
Disallow: /products
- Read your context and instructions
- Compiled the seo audits
- Generated the UI component
SEO audit: /products/[slug]
Framework and rendering: Next.js 14, App Router. That comes from what you told me. You didn't paste a package.json, so I couldn't check the version. app/products/[slug]/page.tsx is an async Server Component with no generateStaticParams and no revalidate, so Next renders it on the server when it's requested. The product name, description and price are in the first HTML response, so Google can read the markup. What stops it is everything around the markup.
What I audited:
| File | Status |
|---|---|
app/products/[slug]/page.tsx | pasted, 11 lines |
app/layout.tsx | only its metadata export, which you described |
public/robots.txt | pasted, 2 lines |
| sitemap | none (no app/sitemap.ts, no public/sitemap.xml) |
| not provided | next.config.js, getProduct, any other routes, the <html> element in the layout |
Line numbers in page.tsx and robots.txt count from the first line you pasted. For layout.tsx you described one statement, not the whole file, so that row names the export instead of a line number. The missing sitemap has no line to point to.
Counts: P0 Indexability: 4 · P1 On-Page Signals: 4 · P2 Enhancements: 3 · Total: 11
| # | File | Line | Issue | Severity |
|---|---|---|---|---|
| 1 | public/robots.txt | L2 | Disallow: /products blocks crawling of every product page | P0 |
| 2 | sitemap (absent) | - | No sitemap, and robots.txt has no Sitemap: line, so Google can only find product URLs through links | P0 |
| 3 | app/products/[slug]/page.tsx | L2 | No not-found handling. An unknown slug either throws on p.name (a 500) or renders an empty 200 (a soft 404) | P0 |
| 4 | app/products/[slug]/page.tsx + app/layout.tsx | L1 / metadata export | No canonical and no metadataBase. ?utm=, ?ref= and slug-case variants all index as separate URLs | P0 |
| 5 | app/products/[slug]/page.tsx + app/layout.tsx | L1 / metadata export | No generateMetadata, so every product's <title> is "My Shop": duplicate titles across the whole catalogue | P1 |
| 6 | app/products/[slug]/page.tsx | L1 | No meta description, so Google writes its own snippet | P1 |
| 7 | app/products/[slug]/page.tsx | L5 | The page has no <h1>. The product name sits in a <div className="title"> | P1 |
| 8 | app/products/[slug]/page.tsx | L6 | <img> has no alt, so the product image means nothing to search or image search | P1 |
| 9 | app/products/[slug]/page.tsx | L1 | No Open Graph or Twitter tags, so shared links show "My Shop" with no image | P2 |
| 10 | app/products/[slug]/page.tsx | L8 | Price is plain text with no Product JSON-LD, so the page can't get price rich results or merchant listings | P2 |
| 11 | app/products/[slug]/page.tsx | L6 | Plain <img> with no width or height. It causes layout shift (CLS), and the likely LCP element gets no priority hint | P2 |
Fix plan, P0: Indexability (no edits made)
P0-1. public/robots.txt: stop blocking /products
Before
User-agent: *
Disallow: /products
After
User-agent: *
Allow: /
Sitemap: https://YOUR-DOMAIN/sitemap.xml
Check this first: if the Disallow is there on purpose to keep a staging site out of the index, it is doing that for production too. In that case the right fix is an environment-aware app/robots.ts and deleting public/robots.txt. Next refuses to build with both. Tell me under
modify and I'll draft that version instead.
P0-2. New file app/sitemap.ts
Before: (file does not exist)
After
import type { MetadataRoute } from 'next';
import { getAllProducts } from '@/lib/products'; // assumption: you have, or can add, a function that lists products
const BASE_URL = 'https://YOUR-DOMAIN';
export default async function sitemap(): Promise<MetadataRoute.Sitemap> {
const products = await getAllProducts();
return [
{ url: BASE_URL, lastModified: new Date() },
...products.map((p) => ({
url: `${BASE_URL}/products/${p.slug}`,
lastModified: p.updatedAt, // drop this line if products have no updated timestamp
})),
];
}
A sitemap file is limited to 50,000 URLs. If the catalogue is bigger, this becomes generateSitemaps.
P0-3. app/products/[slug]/page.tsx L2: return a real 404
Before
export default async function Page({ params }) {
const p = await getProduct(params.slug);
return (
After
import { cache } from 'react';
import { notFound } from 'next/navigation';
const loadProduct = cache(getProduct); // one lookup shared by generateMetadata and Page
export default async function Page({ params }: { params: { slug: string } }) {
const p = await loadProduct(params.slug);
if (!p) notFound();
return (
This assumes getProduct returns null or undefined for an unknown slug. If it throws instead, the call goes in a try/catch that calls notFound().
P0-4. Canonical plus metadataBase
app/layout.tsx, before:
export const metadata = { title: 'My Shop' }
After
import type { Metadata } from 'next';
export const metadata: Metadata = {
metadataBase: new URL('https://YOUR-DOMAIN'),
title: 'My Shop',
};
app/products/[slug]/page.tsx, before (L1, nothing above the component):
export default async function Page({ params }) {
After
import type { Metadata } from 'next';
export async function generateMetadata(
{ params }: { params: { slug: string } }
): Promise<Metadata> {
const p = await loadProduct(params.slug);
if (!p) return {};
return {
alternates: { canonical: `/products/${p.slug}` },
};
}
export default async function Page({ params }: { params: { slug: string } }) {
The canonical comes from the product's stored slug, not the URL. That way /products/Blue-Mug?ref=x points at /products/blue-mug.
Fix plan, P1: On-Page Signals (no edits made)
P1-5. Unique title for each product
app/layout.tsx, before (as P0-4 leaves it):
title: 'My Shop',
After
title: { default: 'My Shop', template: '%s | My Shop' },
app/products/[slug]/page.tsx, before (the generateMetadata return from P0-4):
return {
alternates: { canonical: `/products/${p.slug}` },
};
After
return {
title: p.name, // renders as "Blue Mug | My Shop"
alternates: { canonical: `/products/${p.slug}` },
};
P1-6. Meta description
Before: the same return as above, with no description.
After
const description = p.description.length > 155
? p.description.slice(0, 152).trimEnd() + '...'
: p.description;
return {
title: p.name,
description,
alternates: { canonical: `/products/${p.slug}` },
};
If p.description holds HTML or Markdown, it needs stripping first. Tell me which.
P1-7. L5: product name as the <h1>
Before
<div className="title">{p.name}</div>
After
<h1 className="title">{p.name}</h1>
An <h1> comes with browser-default margins and font size that a <div> doesn't have, so check .title still looks the same.
P1-8. L6: alt text
Before
<img src={p.image} />
After
<img src={p.image} alt={p.name} />
P2-11 replaces this line again with next/image. If you approve both batches, only the P2 version ships.
Fix plan, P2: Enhancements (no edits made)
P2-9. Open Graph and Twitter card
Before: the generateMetadata return as P1 leaves it.
After
return {
title: p.name,
description,
alternates: { canonical: `/products/${p.slug}` },
openGraph: {
title: p.name,
description,
url: `/products/${p.slug}`,
images: [{ url: p.image, alt: p.name }],
type: 'website',
},
twitter: { card: 'summary_large_image' },
};
metadataBase from P0-4 turns a relative p.image into the absolute URL that social previews need.
P2-10. Product JSON-LD (schema.org type: Product with a nested Offer)
Before (L7-L9):
<p>{p.description}</p>
<span>{p.price} EUR</span>
</main>
After
<p>{p.description}</p>
<span>{p.price} EUR</span>
<script
type="application/ld+json"
dangerouslySetInnerHTML={{
__html: JSON.stringify({
'@context': 'https://schema.org',
'@type': 'Product',
name: p.name,
image: [new URL(p.image, 'https://YOUR-DOMAIN').href],
description: p.description,
offers: {
'@type': 'Offer',
price: p.price,
priceCurrency: 'EUR',
url: `https://YOUR-DOMAIN/products/${p.slug}`,
},
}).replace(/</g, '\\u003c'),
}}
/>
</main>
Two things I can't see in what you pasted. First, price has to be a plain decimal like 19.99. If p.price is stored in cents or as a formatted string, it needs converting. Second, Google's merchant listings also want offers.availability, and I've left it out rather than hard-code InStock. If the product has a stock field, tell me its name and I'll wire it in.
P2-11. L6: next/image for CLS and LCP
Before (as P1-8 leaves it):
<img src={p.image} alt={p.name} />
After
import Image from 'next/image';
// ...
<Image
src={p.image}
alt={p.name}
width={800} // replace with the real intrinsic size, or p.imageWidth / p.imageHeight if stored
height={800}
priority // the product image is very likely this page's LCP element
sizes="(max-width: 768px) 100vw, 50vw"
/>
If images come from another host, that host has to be listed in images.remotePatterns in next.config.js, or the build fails.
Once a batch is approved I apply exactly that batch and show you the per-file diff. Then I re-check the patched code against the rows above, print the robots.txt and sitemap diff and the JSON-LD property list, and finish with a found/fixed/skipped/deferred summary and a checklist for after you deploy.
Approve P0 batch? (yes / skip / modify)
If yes, give me your production domain. It goes into robots.txt, the sitemap and metadataBase. Without it I'll leave YOUR-DOMAIN placeholders in and flag every one.
page.tsx
TSX · React component
Example file from a real run - the skill writes it into your workspace.
Subscription · always the latest version
- Secure checkout via Stripe
- Cancel anytime
- Delivered live via MCP
About this skill
Audits a frontend codebase for SEO problems and fixes them in reviewed batches, so nothing about indexing changes without you seeing it first. It finds the framework and rendering mode, then checks what search engines actually get: robots rules, sitemap, status codes for missing pages, titles and descriptions, canonicals, headings, image alt text, Open Graph tags, JSON-LD structured data, hreflang, and the Core Web Vitals issues that are visible in code.
What it does
You get an inventory of every issue with file, line and severity, grouped as P0 (stops indexing), P1 (on-page signals) and P2 (enhancements), and the exact change proposed for each. It then asks to approve the P0 batch, applies only that batch with a per-file diff, re-checks the patched files, and moves on to P1 and P2 the same way. It finishes with what was found, fixed, skipped and deferred, plus a checklist for after deploy.
Try it first
Say "audit my Next.js project for SEO before shipping", or paste a page, its layout and your robots.txt. Check that the first batch it proposes is the one that blocks indexing.
What it needs
Read access to the source, and write access for the batches you approve. Works with Next.js, Nuxt, Remix, Astro, SvelteKit and plain HTML. No network access and no API keys: it audits code, not a live crawl.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Marketing & SEO.
It reads code, it doesn't crawl the live site, so redirects, server headers and anything a CDN changes are out of view unless you paste them. It can't see Search Console or your rankings; Lighthouse and field metrics are only inferred from what the code does. Domain-specific values (production URL, brand name) are left as clear placeholders until you give them. Edits happen only batch by batch, after you approve each one.
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
Free to use
No account required to browse
Trust & safety
Security scan
Security scan passed 28 days ago