seo-audit

name	seo-audit
description	Code-generation skill for diagnosing SEO problems in this repo and producing a repo-specific implementation brief or fix list. Use only when the task should lead to metadata, routing, schema, rendering, or internal-linking code changes.
metadata	{"family":"seo","owner":"seo + engineering","last_reviewed":"2026-04-30T00:00:00.000Z","version":"1.0.0","source_version":"2.0.0"}

SEO Audit

This is a code-generation skill. Use it for diagnosis that directly feeds repo implementation work.

Use When

• rankings, indexation, or discoverability appear weak
• a page type may be missing metadata, canonicals, schema, or links
• the user asks for an SEO audit that should end in code changes
• you need to identify the highest-value SEO fixes before implementing them

Do Not Use When

• the task is already clearly “implement metadata/schema/canonical changes” Then use seo-page-rules or schema-markup.
• the task is primarily AI citation/extractability work Then use ai-seo.
• the task is mostly URL hierarchy or navigation design Then use site-architecture.

First Files To Inspect

• nextjs-app/AGENTS.md
• src/lib/seo/metadata.ts
• src/lib/site/root-metadata.ts
• src/shared/seo/JsonLd.tsx
• src/shared/sitemap/*
• relevant src/app/**/page.tsx

Audit Order

1. crawlability and rendering
2. metadata and canonicals
3. structured data
4. sitemap and robots coverage
5. internal linking and orphan risk
6. page-type completeness

When the audit needs SERP context (who is winning the keyword, what their pages look like), pull data via the semrush-research skill — recipe competitor-snapshot.md for one competitor, keyword-gap.md for a missed-cluster check. Do not guess SERP data.

Repo-Specific Checks

Rendering and Crawlability

• confirm the important content is server-rendered HTML
• confirm critical pages are not hidden behind client-only shells
• confirm dynamic routes use the expected static/rendering pattern when applicable

Metadata

• every public page exports static metadata or generateMetadata
• titles and descriptions are specific, not duplicated boilerplate
• canonicals point to the production URL shape

Structured Data

• use shared schema helpers instead of ad hoc JSON-LD duplication
• confirm the chosen schema matches the page type

Sitemap and Robots

• important routes are represented in shared sitemap logic
• no important pages are accidentally blocked or noindexed

Internal Linking

• important pages are reachable through hub/spoke or related-content patterns
• no obvious orphan routes

Schema Detection Limitation

Do not report “no schema exists” from static fetch alone.

To confirm JSON-LD reliably, use:

1. rendered DOM inspection
2. Google Rich Results Test
3. repo code inspection of schema components

Required Output

Return an SEO Audit Brief with:

1. scope
2. top_issues
3. likely_root_causes
4. owning_files_or_layers
5. recommended_skill_handoff
6. verification_commands

Handoff Rule

• if the problem is implementation-heavy, hand off to seo-page-rules
• if the problem is schema-specific, hand off to schema-markup
• if the problem is IA/internal-linking, hand off to site-architecture