JSON-LD over Microdata and RDFa
The skill settles the format first. JSON-LD sits in its own script tag, so you can add, change, or remove it without touching the visible HTML, and Google recommends it. Microdata is woven into your markup, which makes it harder to audit, and the skill sees little reason to start anything new in RDFa.
It prefers the page head, and it accepts the body when a content management system leaves no other option.
Schema types by kind of site
- A software company gets Organization on the homepage, SoftwareApplication on product pages, FAQPage on pricing and support pages, and BlogPosting on posts.
- An online store gets Organization, then Product with Offers and AggregateRating on every product page, plus BreadcrumbList on category and product pages.
- A local services business gets LocalBusiness on the homepage and every location page, FAQPage on service pages, and Review where reviews are published.
- A publisher gets Organization, Article or BlogPosting on each article, and BreadcrumbList across the site.
The highest priority types go in first, get validated, and only then does the next set follow.
Rules for every page
Markup describes only what a visitor can see on the page. Reviews written by your own team never appear as independent ratings. Product markup without an Offers block cannot show a price, so the skill always includes one, and a rating needs a numeric value before stars can appear.
Each location gets its own LocalBusiness block with its own address, phone, and hours. Category pages never receive Product markup, because a list of products is not a product.
Validation before it ships
The Rich Results Test comes first. It shows which result types a page qualifies for, which properties it detected, and which required ones are missing. The skill reads a warning as a note and an error as a blocker. The Schema Markup Validator then catches broken syntax, and Search Console’s Enhancements reports show errors across many pages at once.
After a fix, the skill has you request reindexing and allow 48 to 72 hours before judging the result.
Examples and rules in the schema file
Ready to adapt JSON-LD samples for Organization and FAQPage live in the schema skill file, along with notes for WordPress, Shopify, Webflow, and Next.js. The repository is released under the MIT License, “Copyright (c) 2026 ScayverGraphix.” The skills follow the Agent Skills spec published by vercel-labs/skills.
Point it at one template.
Mention JSON-LD, rich results, or a specific type such as Product schema, and the agent loads the skill. The project readme lists it after seo-content-strategy and seo-blog-writer on the path for building search traffic. You can add the three at once with npx skills add scayver/marketing-skills --skill seo-content-strategy seo-blog-writer schema.
Markup is a prerequisite, not a promise.
The skill says so itself. Structured data makes a page eligible for rich results and easier for AI systems to read correctly, but neither outcome is certain. It also needs upkeep: an old price, a past event, or a stale modified date in the markup is an error the moment the page changes.
A larger technical problem behind the markup
Markup errors can sit beside other problems, such as a blocked route or two pages that declare conflicting canonical destinations. Our Technical SEO service traces a problem to the affected routes, defines the change, and checks what happened after release.