seo

Step 6 of the pipeline (WORKFLOW.md), complementing vue-nuxt-vuetify-conventions/react-nextjs-conventions: applies only to pages meant to be indexed by a search engine (not to back-offices, authenticated internal apps, or dashboards). The rules follow the web standards and the search engines' published guidance, not any assistant or tool.

When

As soon as a frontend page is public and has to be findable through search: during code (6) or at review time (review, 8) if the diff touches public pages.

Steps

Read the sections the task touches. One file per section under references/; a single public page diff is usually §1 to §4, a launch or a migration adds §5, and a page with structured data, a published date or a share card adds §6; an assistant-visibility audit is §8.

§	Covers	Read it when	File
1	Meta and indexing: title, description, canonical, robots, social tags, hreflang	any public page	01-meta-indexing.md
2	HTML semantics a crawler and a screen reader share	headings, links, images, or content reaching the HTML	02-html-semantics.md
3	Core Web Vitals	the largest element, image sizing, main interactions	03-core-web-vitals.md
4	Structured data, sitemap and URL shape at the page level	a page type could carry structured data, or sitemap, robots or URLs change	04-structured-sitemap-urls.md
5	Crawl and index health of a built site: statuses, redirects, canonical and directive conflicts, sitemap, link graph, URL form	before a launch or migration, a URL change, or a site audit	05-crawl-health.md
6	Structured data types, dates and social previews	JSON-LD is added, a date or an author is published, or a page is meant to be shared	06-structured-data-social.md
7	Content trust signals and local presence	a site publishes advice or articles, a sensitive subject, or a physical location	07-content-trust-local.md
8	AI-assistant visibility audit: crawler access as the owner's decision, renderability, markup that matches the page, facts that agree everywhere, listings, repeat snapshot tests with no score, no promised ranking, no instructions to assistants in a page	the question is whether assistants can reach, read and quote a public site correctly	08-ai-assistant-visibility.md

How a robots.txt is actually matched, and the sitemap size limits, are in skills/seo-crawl-control-pitfalls; finding the cause of a poor vital is skills/cwv-diagnosis-pitfalls, and §3 here stays the page rules.

Output / checkpoint

The sections reviewed on the diff touched; for a broader audit of a site already in production (not just the diff in progress), see the keymaker agent.

Guardrails

• Never applies to non-public pages (auth, back-office, internal dashboard): don't impose this checklist outside its scope.
• No JSON-LD over-engineering: only the content types that get a real benefit from it (product page, article), not a systematic addition.
• Never promise a ranking or a recommendation by an assistant, report a visibility score, or write instructions to assistants into a page; record what an assistant answered, with the date (§8).
• This block has no dedicated in-house production experience yet: to be confronted with the first real SEO audit, not to be treated as proven doctrine.

Origin

Provenance, the re-checks and the correction log are in references/origin.md.