seo-config-audit

name	seo-config-audit
description	Audit Next.js App Router SEO config across robots, sitemap, page-level Metadata, OG, Twitter, canonicals, and JSON-LD — catch contradictions like sitemap vs robots disallow
source	auto-skill
extracted_at	2026-07-15T10:37:00.000Z

Task: Audit SEO configuration consistency

In a Next.js App Router project, verify that all SEO-related files are aligned. Follow the four-layer check below in order — each layer can reveal issues missed by the others.

Layer 1: robots.ts — What crawlers are allowed / blocked?

Read apps/web/src/app/robots.ts. Check:

• rules entry for userAgent: '*' — the global allow/disallow rules.
• rules entry for userAgent: AI_BOTS (or the actual AI bot array name) — the AI-bots-specific rules.
• Ensure /api is disallowed (should never be crawled).
• Client-only shell routes (e.g. /studio) must appear in the AI-bot disallow list — if crawlable content lives elsewhere, these routes have nothing for search engines to index.

Layer 2: sitemap.ts — What URLs do you advertise?

Read apps/web/src/app/sitemap.ts. Check:

• Static routes: root, /studio, /templates, /blog, etc.
• Dynamic routes: getAllTemplateSlugs(), getAllPostSlugs() — verify these functions exist and actually return slugs.
• Every route in the sitemap must be crawlable per robots.ts. If a route is disallowed for AI bots in robots.ts, remove it from the sitemap — contradictory signals hurt indexing.
• For static pages, lastModified can be new Date() (fine for SSG). Content pages should use the post's actual updatedAt if available.

Layer 3: Page-level Metadata — Per-page SEO signals

For each important page file, check:

File	Required Metadata fields
page.tsx (landing, /templates, /blog)	title, description, alternates.canonical, openGraph.{title, description, url, type}
[slug]/page.tsx (template, blog detail)	generateMetadata() returning title, description, keywords, alternates.canonical, openGraph.{title, description, url, type: 'article'}, twitter.{card, title, description}

• openGraph.url should be the relative path (e.g. '/templates/slug'), not the full URL.
• type should be 'article' for blog posts, 'website' for landing/hub pages, 'website' for template listings.
• twitter.card should be 'summary_large_image' for article/detail pages.

Layer 4: JSON-LD — Structured data

Check each page that renders rich content:

Page	JSON-LD to emit
Template detail ([slug]/page.tsx)	techArticleLd or similar — @type: 'TechArticle' with author, datePublished, headline
Blog post ([slug]/page.tsx)	articleLd + faqLd (if FAQs exist) + breadcrumbLd
Template listing	Optional — ItemList schema if listing many items
Root layout.tsx	Site-wide SoftwareApplication + Organization + WebSite entity graph

Catching contradictions

The most common SEO issues are conflicting signals between layers. Here are the patterns to catch:

1. Sitemap vs robots mismatch

   • A URL appears in sitemap.ts but is disallowed for AI bots in robots.ts.
   • Fix: Remove the URL from the sitemap. If it's a client-only shell, it has no crawlable content.

2. Canonical points to wrong page

   • alternates.canonical on a sub-page points to a different route instead of itself.
   • Fix: The canonical for /templates/slug should be /templates/slug, not /.

3. Robots allow index but OG is missing

   • Page has robots: { index: true } but no openGraph or twitter metadata.
   • Fix: Add OG + Twitter cards; social sharing will fall back to bare title/description but look unpolished.

4. Type mismatch in OG

   • Blog detail uses type: 'website' instead of 'article'.
   • Fix: Use 'article' for blog posts / long-form content; 'website' for landing/hub pages.

Checklist for any new page addition

When adding a new content page, verify:

• Sitemap includes it (static or dynamic via a getSlugs() function)
• robots.ts permits crawling for it (check both '*' and AI-bot rules)
• Page-level Metadata object has title, description, alternates.canonical
• openGraph with title, description, url, type
• twitter with card, title, description
• JSON-LD structured data if the page renders rich content
• No contradictory noindex on the page if it should be indexed