json-ldlisted
Install: claude install-skill Mavens-Tech-Lab/skills
# JSON-LD Structured Data
Implement and repair schema.org structured data (as JSON-LD) grounded in three sources of truth:
1. **The codebase's data models** — what entities actually exist and where they render
2. **The live vocabulary** — the current schema.org release and Google's current rich-results requirements, fetched fresh every run
3. **The real data flow** — every emitted property traced to a database column, CMS field, or frontmatter key; never hardcoded, never invented
The workflow is: **entity discovery scan → fetch fresh definitions → map data flow → one intake round with the user → implement in parallel → validate against rendered HTML**.
## Critical framing (read before emitting a single property)
- **Fresh definitions or stop.** schema.org ships new releases continuously (v30.0 landed 2026-03; types get added, superseded, and graduated from pending) and Google adds and *retires* rich-result types without ceremony (FAQ and HowTo rich results were demoted in 2023). Model memory about either is stale by construction. Every type and every property you emit MUST be verified against a definition fetched this run (`references/fresh-definitions.md`). If network access is unavailable, tell the user and stop — do not proceed from memory unless they explicitly accept stale definitions, and mark the output as unverified if they do.
- **Real data only.** Every value comes from a traced source: DB column, CMS field, frontmatter key, site config, or an intake answer. A