URL Metadata & Schema.org Extractor – Open Graph, Twitter Card, JSON-LD, Logo

Give it a list of URLs and get back everything a link preview, CRM or AI agent needs to know about each page: title, meta description, canonical, language, robots, all Open Graph tags, Twitter card, JSON-LD / Schema.org (with @graph flattened), microdata, favicon, apple-touch-icon and the site logo, theme colour, hreflang alternates, oEmbed and RSS/Atom feeds, social profile links, site name, author and publish dates. Every row also gets a list of missing or invalid tags, so the same run doubles as an SEO / social-sharing check.

Plain HTTP, no browser: $1 per 1,000 URLs, failed URLs are free.

Related Actors (website toolkit): Sitemap Extractor to get every URL of a site first, Website to Markdown for the page content as clean Markdown, SEO Audit Crawler for a full-site crawl with broken links and redirects, Website Contact Finder for emails and phones, and Tech Stack Detector for the software a site runs on.

What it extracts per URL

• Basics: final URL after redirects, HTTP status, response time, <title>, meta description, canonical, <html lang>, meta robots, X-Robots-Tag, viewport, charset, keywords, H1s.
• Open Graph: every og:*, article:*, product:*, fb:* tag (repeated tags become arrays), plus a structured ogImages list with width, height, alt and secure URL.
• Twitter / X card: every twitter:* tag.
• JSON-LD: every block parsed (lenient: trailing commas, HTML comments and CDATA are repaired), @graph flattened, @context kept. jsonLdTypes lists the top-level types and jsonLdAllTypes the nested ones. Ready-made summaries:
  • organization – name, URL, logo, sameAs, email, telephone
  • product – name, brand, SKU, GTIN, image, price, currency, availability, rating
  • article – headline, published/modified dates, authors, image, publisher
  • breadcrumbs – position, name, URL
• Microdata and RDFa (light): itemscope/itemprop items with nested properties, and RDFa typeof types.
• Brand assets: best favicon (falls back to /favicon.ico), appleTouchIcon, every icon with sizes, web manifest, and a best-guess logo with its source: JSON-LD Organization logo → og:logo → microdata → <img> with "logo" in the header/nav → large touch icon. themeColor from <meta name="theme-color">.
• Link preview fields: previewTitle, previewDescription, image, siteName – the same fallbacks Slack, LinkedIn and iMessage use (OG → Twitter → HTML).
• Image check (optional, on by default): reads the first 64 KB of the preview image and the logo and reports HTTP status, content type, file size and pixel width × height (PNG, JPEG, GIF, WebP, AVIF, ICO, SVG).
• Discovery links: hreflang alternates, oEmbed endpoints, RSS/Atom/JSON feeds.
• Social profiles: from JSON-LD sameAs and header/footer links – Facebook, X/Twitter, Instagram, LinkedIn, YouTube, TikTok, Pinterest, GitHub, Threads, Bluesky, Mastodon, Discord, Telegram and more. Share buttons and deep links (posts, repos) are filtered out.
• People and dates: authors (meta, article:author, JSON-LD), author URL, published and modified times.

Missing / invalid tag checks

Each row has issues (code, severity, message), issueCodes, errorCount and warningCount. Checks include:

Severity	Codes
error	missing-title, multiple-canonicals, og-image-not-absolute, invalid-json-ld, invalid-hreflang, preview-image-broken
warning	title-too-long (>60), title-too-short, missing-meta-description, missing-canonical, missing-html-lang, missing-viewport, noindex, noindex-header, missing-og-title, missing-og-image, og-image-too-small / preview-image-too-small (<200 px), invalid-twitter-card, json-ld-missing-type, json-ld-missing-context, hreflang-missing-self-reference, logo-broken
info	meta-description-too-long (>160) / -too-short, canonical-points-elsewhere, missing-og-description / -url / -type, missing-twitter-card, no-structured-data, product-without-offer-price, missing-favicon-link, missing-apple-touch-icon, hreflang-missing-x-default, nofollow

Use cases

• Link previews / unfurling in your chat app, CMS, bookmarking tool or newsletter editor.
• CRM and lead enrichment: company logo, favicon, brand colour, description and social profiles from a website URL.
• AI agents: a cheap "what is this page?" tool that returns structured facts instead of raw HTML.
• SEO and social-sharing QA: find pages without OG images, broken or tiny share images, invalid JSON-LD, missing canonicals or hreflang errors – combine with the Sitemap Extractor to check a whole site.
• E-commerce monitoring: product name, price, currency and availability from JSON-LD on any shop.
• Content research: article headline, authors and publish dates from news sites and blogs.

Input

Field	Description
urls	Page URLs (bare domains work). One row each
probeImages	Check the preview image and logo: status, size, dimensions (default on)
includeJsonLd	Output full JSON-LD objects (default on; summaries are always included)
includeMicrodata	Also read microdata and RDFa (default on)
maxUrls, respectRobotsTxt, maxConcurrency, maxRequestsPerHost, requestTimeoutSecs, maxRetries, userAgent, proxyConfiguration	Advanced

{
  "urls": ["https://stripe.com", "https://www.allbirds.com/products/mens-wool-runners", "https://paulgraham.com/greatwork.html"],
  "probeImages": true
}

Output

The Output tab has four tables – Link preview & brand, SEO tag check, Structured data and Social profiles, feeds, hreflang, oEmbed – plus the full JSON and a run SUMMARY. A real row (stripe.com, shortened):

{
  "url": "https://stripe.com",
  "finalUrl": "https://stripe.com",
  "status": 200,
  "ok": true,
  "title": "Stripe | Financial Infrastructure to Grow Your Revenue",
  "siteName": "Stripe",
  "previewTitle": "Stripe | Financial Infrastructure to Grow Your Revenue",
  "image": "https://b.stripecdn.com/mkt-ssr-statics/assets/_next/static/media/stripe.37119086.jpg",
  "imageProbe": {"status": 206, "ok": true, "contentType": "image/jpeg", "bytes": 312818, "format": "jpeg", "width": 2048, "height": 1024},
  "logo": "https://images.stripeassets.com/fzn2n1nzq965/1hgcBNd12BfT9VLgbId7By/01d91920114b124fb4cf6d448f9f06eb/favicon.svg",
  "logoSource": "json-ld",
  "favicon": "https://assets.stripeassets.com/fzn2n1nzq965/nu0iR4KqmkdJ52fgRts09/c3a10df318e7ffb1f516073daadea625/favicon.ico",
  "jsonLdTypes": ["WebSite", "Organization"],
  "socialProfiles": [
    {"platform": "x", "url": "https://twitter.com/stripe", "source": "json-ld sameAs"},
    {"platform": "linkedin", "url": "https://www.linkedin.com/company/stripe", "source": "json-ld sameAs"}
  ],
  "issueCodes": [],
  "errorCount": 0,
  "warningCount": 0
}

A product page adds product ({"name": "Men's Wool Runner", "brand": "Allbirds", "price": "110", "priceCurrency": "USD", "availability": "InStock", ...}); an article adds article, authors and publishedTime. A page without any Open Graph (paulgraham.com) still returns its title and favicon, with 12 issue codes such as missing-og-image and missing-meta-description. Failed URLs are rows with ok: false, errorType (network, http, robots, not-html, no-metadata) and error. See SAMPLE_OUTPUT.json for full real rows.

Pricing

Pay per event:

Event	Price
URL processed (page fetched and parsed)	$0.001 ($1 per 1,000 URLs)
Actor start	$0.00005
Failed URLs (DNS, timeout, 4xx/5xx, robots.txt, non-HTML, bot challenge)	free

Compared with other metadata Actors in the Apify Store (listed prices, October 2026):

Actor	Listed price
autofacts/metadata-scraper	$5 / 1,000 URLs
njoylab metadata extractor	$10 / 1,000 URLs
This Actor	$1 / 1,000 URLs, image checks included

Set a maximum cost per run in the run options: the Actor processes as many URLs as fit, never fetches a URL it cannot bill, and stops cleanly.

FAQ

Does it run JavaScript? No. It reads the HTML the server sends, which is exactly what Facebook, LinkedIn, Slack and X read to build link previews – so if a tag is missing here, it is missing for them too. Pure JavaScript apps without server-rendered meta tags come back as no-metadata and are not charged.

Some sites return 403 or time out. A few large sites (e.g. Etsy, NYTimes, The Guardian, NPR) block automated clients at their CDN. The Actor sends an honest bot User-Agent with a contact address and respects robots.txt, so these rows fail and are free. You may set your own userAgent or a proxy for sites you are allowed to access.

Does it respect robots.txt? Yes by default, including Crawl-delay; disallowed URLs are reported with errorType: "robots" and not charged. At most 2 parallel requests per host.

How is the logo chosen? In order: JSON-LD Organization/Brand logo, og:logo, microdata logo, an <img> with "logo" in its attributes inside the header/navigation (customer-logo walls are skipped), the image linking to the home page, and finally a touch icon of 120 px or more. logoSource tells you which one was used. Inline SVG logos have no URL and cannot be returned.

Can I get CSV or Excel? Yes, every Apify dataset exports to CSV, Excel, JSON or XML. Use the table views for flat columns.

Use with AI agents / Apify MCP

Connect the Actor through the Apify MCP server (https://mcp.apify.com?actors=gazidev/metadata-schema-extractor) and Claude, ChatGPT, Cursor or any MCP client can call it with prompts like "get the logo, brand colour and social profiles of these 20 company websites" or "check which of our blog posts have no og:image". Over the API: POST https://api.apify.com/v2/acts/gazidev~metadata-schema-extractor/run-sync-get-dataset-items with {"urls": ["https://example.com"]}. Turn includeJsonLd off for smaller responses in agent contexts.

Categories

SEO tools · Developer tools · AI · Lead generation