@jdevalk/seo-graph-core

Pure schema.org JSON-LD graph builders. Runtime-agnostic core for agent-ready SEO.

What this is

A small, dependency-light library that builds a valid schema.org @graph from a set of typed inputs. It does one thing: turn structured page data into byte-correct JSON-LD that search engines and agents can consume.

It does not know anything about Astro, Next.js, EmDash, WordPress, or any other runtime. Use @jdevalk/astro-seo-graph for the Astro integration, or consume this directly from your own CMS or framework.

For detailed usage — including all builder signatures, site-type recipes, and schema.org best practices — see AGENTS.md.

Install

npm install @jdevalk/seo-graph-core

What you get

Graph assembly

IndexNow

Piece builders

All builders take an input object and the IdFactory, and return a plain object with @type and @id. Builders for CreativeWork subtypes (WebSite, WebPage, Article) share a common set of optional fields via CreativeWorkFields: description, inLanguage, datePublished, dateModified, about, copyrightHolder, copyrightYear, copyrightNotice, license, and isAccessibleForFree.

All schema.org properties are accepted at the top level with full autocomplete from schema-dts. Dedicated builders handle ID generation, date conversion, and non-trivial transforms. Use buildPiece<Type> for everything else (Person, Organization, Blog, Product, Recipe, Event, etc.).

Usage

import {
    makeIds,
    assembleGraph,
    buildWebSite,
    buildArticle,
    buildWebPage,
    buildBreadcrumbList,
} from '@jdevalk/seo-graph-core';

const ids = makeIds({ siteUrl: 'https://example.com' });
const url = 'https://example.com/my-post/';

const graph = assembleGraph([
    buildWebSite(
        {
            url: 'https://example.com/',
            name: 'Example',
            publisher: { '@id': ids.person },
            inLanguage: 'en-US',
        },
        ids,
    ),
    buildWebPage(
        {
            url,
            name: 'My Post',
            isPartOf: { '@id': ids.website },
            breadcrumb: { '@id': ids.breadcrumb(url) },
            datePublished: new Date('2026-04-07'),
        },
        ids,
    ),
    buildArticle(
        {
            url,
            isPartOf: { '@id': ids.webPage(url) },
            author: { '@id': ids.person },
            publisher: { '@id': ids.person },
            headline: 'My Post',
            description: 'A post about something interesting.',
            datePublished: new Date('2026-04-07'),
        },
        ids,
        'BlogPosting',
    ),
    buildBreadcrumbList(
        {
            url,
            items: [
                { name: 'Home', url: 'https://example.com/' },
                { name: 'My Post', url },
            ],
        },
        ids,
    ),
]);

// graph === { '@context': 'https://schema.org', '@graph': [...] }

IndexNow

Runtime-agnostic helpers for the IndexNow protocol. Submit URLs to participating search engines (Bing, Yandex, Seznam, Naver, Yep) through the neutral aggregator at api.indexnow.org.

import {
    generateIndexNowKey,
    getIndexNowKeyFileContent,
    submitToIndexNow,
    validateIndexNowKey,
} from '@jdevalk/seo-graph-core';

const key = generateIndexNowKey(); // 32-char hex, persist this

// Serve this body at https://example.com/<key>.txt
const keyFileBody = getIndexNowKeyFileContent(key);

const results = await submitToIndexNow({
    host: 'example.com',
    key,
    urls: ['https://example.com/blog/new-post/'],
});

URLs not on host and duplicates are filtered automatically. Bulk submissions are chunked at 10,000 URLs per request. submitToIndexNow never throws on network errors — it returns one result per chunk with ok, status, and message. Pass endpoint to override the default aggregator, keyLocation to point at a non-default key-file path, or fetch for testing.

Why

Read more about why this project exists.

License

MIT © Joost de Valk