schema-maker

<objective> Turn a finished markdown post into the structured data that decides whether an answer engine can quote it. Deterministic: reads frontmatter and headings, emits validated JSON-LD. No model call, no network. </objective> ## ROLE Structured data used to be about rich results in Google. That era ended on 2026-05-07, when Google retired the FAQ rich result. The markup did not stop mattering — it changed job. AI Overviews, ChatGPT browsing, Perplexity and Gemini parse `FAQPage` first when deciding whose answer to quote, and pages carrying Tier 1 schema turn up in AI summaries markedly more often than pages without it. So this skill optimises for citation, not for a SERP widget. Not for writing the copy — that is `landing-copy` (marketing surfaces) or `essay-write` (longform). Not for checking whether the prose is extractable — that is `writer --aeo`. This skill assumes the text exists and describes it. ## PIPELINE 1. **Read the post.** Frontmatter supplies title, date, description and tags. The body supplies the FAQ pairs. 2. **Extract Q&A automatically.** Every question-form `##` / `###` heading plus the first paragraph under it becomes a `Question` / `acceptedAnswer` pair. This is why question headings are worth the trouble twice over: they help a human scan, and they are what an engine lifts. Run `writer --aeo` first if the headings are still statements. 3. **Attach the author entity.** `knowsAbout` is the property that ties an author to a t...