Astro — SEO & JSON-LD

Every field you need for head tags and structured data is already on entry.data — the loader surfaces the full article. You emit the tags; Ghost Writr doesn’t inject anything into the page.

Head tags

Set the title, description, and canonical from entry.data. seoTitle falls back to title server-side, so it’s always a string; the rest can be null, so guard them.

src/pages/blog/[slug].astro (head)

---
const { seoTitle, seoDescription, canonicalUrl, image } = p.data;
const canonical = canonicalUrl ?? new URL(p.id, Astro.site).href;
---
<title>{seoTitle}</title>
<meta name="description" content={seoDescription ?? ""} />
<link rel="canonical" href={canonical} />
<meta property="og:title" content={seoTitle} />
{seoDescription && <meta property="og:description" content={seoDescription} />}
{image && <meta property="og:image" content={image.url} />}
{image?.width && <meta property="og:image:width" content={String(image.width)} />}

canonicalUrl is the article’s declared canonical (set when the content is syndicated from elsewhere). When it’s null, the page is its own canonical — fall back to the page URL, as above. This is your call; see SEO & JSON-LD.

JSON-LD

Build the graph in frontmatter and emit it with set:html (it’s trusted, generated server-side from your own data — not user input):

src/pages/blog/[slug].astro (json-ld)

---
const a = p.data;
const url = a.canonicalUrl ?? new URL(p.id, Astro.site).href;

const author =
  a.author.kind === "real"
    ? {
        "@type": "Person",
        name: a.author.name,
        jobTitle: a.author.jobTitle ?? undefined,
        ...(a.author.sameAs.length ? { sameAs: a.author.sameAs } : {}),
      }
    : { "@type": "Organization", name: a.author.name };

const jsonLd = {
  "@context": "https://schema.org",
  "@type": a.formatType, // schema.org type hint, e.g. "Article"
  headline: a.title,
  description: a.seoDescription ?? undefined,
  inLanguage: a.inLanguage,
  datePublished: a.publishedAt.toISOString(),
  dateModified: a.updatedAt.toISOString(),
  mainEntityOfPage: url,
  ...(a.image ? { image: a.image.url } : {}),
  author,
};
---
<script type="application/ld+json" set:html={JSON.stringify(jsonLd)} />

author.kind picks the author type

real → Person, persona → Organization

author.kind is "real" for a named human (emit a Person, with jobTitle and sameAs profile links) and "persona" for a brand byline (emit an Organization). Branching on kind keeps your structured data honest — a brand byline shouldn’t claim to be a person.

Fields at a glance

• formatType → JSON-LD @type (schema.org hint, e.g. "Article").
• inLanguage → BCP-47 inLanguage and an html lang.
• publishedAt / updatedAt → datePublished / dateModified (they’re Dates — call .toISOString()).
• image → og:image and JSON-LD image (with width/height/srcset/alt when present).
• canonicalUrl → <link rel="canonical"> and mainEntityOfPage.

See Collection schema for the full entry.data shape and The article shape for the underlying contract.