agentsclimarketplace

Seo schema builder

Skill gaguero/seo-agent-skills/seo-schema-builder

Curated SEO Agent Skills pack for AI agents (Agent Skills open standard): technical audit, content briefs, JSON-LD schema, GEO/AEO, and anti-spam quality gates.

Install
npx -y skills add gaguero/seo-agent-skills --skill seo-schema-builder

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Generates and validates schema.org structured data as JSON-LD for rich results — Organization, LocalBusiness, Product, Service, Article, Breadcrumb, FAQPage, Event, Recipe, Review/AggregateRating, and more. Enforces strict no-fabrication rules: it refuses to invent ratings, reviews, prices, availability, authors or any value not backed by real, on-page data. Use when someone asks for structured data, JSON-LD, schema markup, rich results / rich snippets, or wants a page eligible for enhanced search appearances.

The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.

SKILL.md

4.6 KB, 949 tokens by cl100k_base, as published. Nobody here has run it

Schema / JSON-LD Builder

Structured data helps search engines understand a page and unlocks rich results. But it is also the single easiest place for an AI agent to fabricate data and earn a manual penalty. This skill's first job is to refuse fake markup.

The non-negotiable rule

Only mark up content that is real, accurate, and visible to the user on the page. If a required value isn't backed by real data, do not invent it — output "data needed" and stop.

Never generate: ratings/review counts you don't have, reviews you didn't receive, prices/availability you can't confirm, authors who don't exist, or markup for content that isn't on the page. Google's structured-data policies prohibit this and it risks manual actions.

Workflow

  1. Identify the page type → pick the matching schema type(s):

    PagePrimary type(s)
    Brand/homeOrganization (+ sameAs, logo)
    Local business / locationLocalBusiness (subtype) + NAP + geo + openingHours
    ProductProduct + Offer (+ AggregateRating/Review only if real)
    ServiceService
    Blog/news/guideArticle / BlogPosting / NewsArticle + author + dates
    Any deep pageBreadcrumbList
    Genuine Q&A on pageFAQPage (see strict rules below)
    EventEvent + dates + location + offers
    Recipe / How-toRecipe / HowTo

  2. Collect real values from the page (or ask). Map each schema property to a visible, verifiable source. If a value isn't on the page and can't be confirmed, either omit the optional property or report it as required-but-missing.

  3. Generate JSON-LD — always <script type="application/ld+json">, placed in <head> or <body>. Prefer JSON-LD over Microdata/RDFa. One graph or multiple scripts both fine; use @id to link nodes (e.g. Article → Organization publisher).

  4. Validate before shipping:

    • Run scripts/validate.py <file-or-url> for structural/required-field checks and fabrication smells.
    • Then the Rich Results Test (https://search.google.com/test/rich-results) for eligibility, and validator.schema.org for vocabulary correctness.

Type-specific guardrails

  • Review / AggregateRating — only if the page genuinely displays real reviews/ ratings collected from real customers. Self-serving reviews of your own business/products are not eligible. No real reviews → no review markup. Period.
  • FAQPage — only for questions-and-answers actually shown on the page, that are genuinely informational. Don't fabricate FAQs to game snippets; don't use FAQPage for ads/promos or a single question. (Google has narrowed FAQ rich-result display — value it for clarity/AEO, not guaranteed snippets.)
  • Product Offer — price, priceCurrency, availability must be current and match the page. Don't guess currency or stock.
  • LocalBusiness — NAP (name, address, phone) must be identical to the site's visible NAP and other citations. Include geo, openingHoursSpecification, areaServed, url, image. Pick the most specific subtype (e.g. Resort, CarRental, TravelAgency).
  • Article — real author (Person/Organization), datePublished, dateModified, headline, image, publisher.

Output

  • The JSON-LD block(s), ready to paste.
  • A line-by-line note of which value came from where (so a human can verify).
  • An explicit list of any required fields that are missing real data, blocking those types until supplied.

Run the result through seo-quality-gates before publishing.

See references/schema-types.md for templates and required fields, and scripts/validate.py for local checks.

What ships with it: 4 files

14.6 KB alongside SKILL.md, 2 of them executable

references/

scripts/

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.