Page Metadata Auditor: SEO, Open Graph, Twitter Cards, JSON-LD

Give it a list of page URLs. For each page you get the metadata search engines and social networks read — title, meta description, canonical, meta robots and X-Robots-Tag, viewport, lang, the first <h1> and the <h1> count, Open Graph, Twitter card, JSON-LD @types and hreflang alternates — plus a plain list of issues with a severity. One flat JSON row per page, ready for a spreadsheet, a CI check, or an AI agent.

It reads the pages you list. It does not crawl, and it does not run JavaScript: metadata that a page only injects client-side is not seen.

What you get per page

Field	Meaning
url, finalUrl	The URL you gave (normalised) and where redirects ended (max 5 redirects)
statusCode, redirectCount, contentType	Final HTTP response
title, titleLength	<title> text (whitespace collapsed, entities decoded) and its character count
description, descriptionLength	<meta name="description"> and its character count
canonical	<link rel="canonical"> href as written
robots, xRobotsTag	<meta name="robots"> and the X-Robots-Tag response header
viewport, lang	<meta name="viewport">, <html lang>
h1, h1Count	Text of the first <h1> and how many there are
openGraph	title, description, image, url, type, siteName, locale
twitter	card, site, creator, title, description, image
jsonLd	count of application/ld+json blocks, invalid (not valid JSON), types (all @type values incl. @graph)
hreflang	[{lang, href}] from <link rel="alternate" hreflang>
bodyTruncated	true when the page was larger than 1 MB and only the first 1 MB was parsed
status	OK (no issues) or ISSUE
issues	List of {severity, code, message}
checkedAt	ISO timestamp

Issue codes (character counts, not pixels):

• error: TITLE_MISSING, JSONLD_INVALID
• warning: TITLE_TOO_LONG (> 60), TITLE_TOO_SHORT (< 10), DESCRIPTION_MISSING, DESCRIPTION_TOO_LONG (> 160), DESCRIPTION_TOO_SHORT (< 50), H1_MISSING, CANONICAL_MISSING, VIEWPORT_MISSING, OG_IMAGE_MISSING, OG_IMAGE_NOT_ABSOLUTE, NOINDEX (meta robots or X-Robots-Tag says noindex/none)
• info: H1_MULTIPLE, CANONICAL_DIFFERENT (canonical points elsewhere; trailing slash and #fragment ignored), LANG_MISSING, OG_TITLE_MISSING, OG_DESCRIPTION_MISSING, TWITTER_CARD_MISSING, JSONLD_MISSING

Error rows (free, no metadata): INVALID_URL, HTTP_ERROR (4xx/5xx, with statusCode), NOT_HTML, UNSUPPORTED_ENCODING (e.g. a zstd body this Actor cannot decode), PARSE_ABORTED (page exceeded the parse time budget), ROBOTS_DISALLOWED, REFUSED_PRIVATE_ADDRESS, TIMEOUT, FETCH_FAILED, CHECK_FAILED.

Example input

{ "urls": ["https://www.python.org/", "example.com"], "maxUrls": 100 }

Example output (real run, trimmed)

{
  "url": "https://www.python.org/",
  "finalUrl": "https://www.python.org/",
  "statusCode": 200,
  "title": "Welcome to Python.org",
  "titleLength": 21,
  "description": "The official home of the Python Programming Language",
  "descriptionLength": 52,
  "canonical": null,
  "viewport": "width=device-width, initial-scale=1.0",
  "lang": "en",
  "h1Count": 5,
  "openGraph": { "title": "Welcome to Python.org", "image": "https://www.python.org/static/opengraph-icon-200x200.png", "type": "website", "siteName": "Python.org" },
  "twitter": { "card": null },
  "jsonLd": { "count": 1, "invalid": 0, "types": ["WebSite"] },
  "hreflang": [],
  "issues": [
    { "severity": "info", "code": "H1_MULTIPLE", "message": "5 <h1> elements" },
    { "severity": "warning", "code": "CANONICAL_MISSING", "message": "No <link rel=\"canonical\">" },
    { "severity": "info", "code": "TWITTER_CARD_MISSING", "message": "No twitter:card" }
  ],
  "status": "ISSUE"
}

Pricing

Pay per event: a small start fee per run plus one page-audited event for each HTML page fetched, parsed and delivered with its metadata. Rows without metadata are free: invalid URLs, HTTP 4xx/5xx pages, non-HTML or undecodable-encoding responses, pages robots.txt disallows, refused private addresses, timeouts and fetch errors. If your spending limit is reached, the run stops cleanly instead of doing unpaid work. Current prices are shown on this page.

Good to know

• Public pages only. IP addresses, localhost, internal hostnames (.local, .internal, …) and URLs with credentials are rejected. Every connection, including each redirect hop, is made to an address validated as public at connect time, and redirects to private address literals (127.0.0.1, ::1, 169.254.169.254, …) are refused.
• robots.txt is honoured. It is read once per origin and checked before every request, including redirect hops; a disallowed page is returned as a free ROBOTS_DISALLOWED row and never requested. A robots.txt answering 5xx is treated as "disallow all"; 4xx as "no restrictions".
• A bare domain (example.com) is audited as https://example.com/. Duplicate URLs are removed and #fragments dropped.
• Limits: maxUrls default 100, hard cap 1000; 5 pages in flight overall and at most 2 requests in flight to one host, with ~100 ms between request starts to the same host; 12 s per request, 30 s per page (a redirect chain cannot run 5 x 12 s); 1 MB of HTML parsed per page (gzip, deflate — zlib-wrapped or raw — and brotli are decoded; the page's charset is honoured). Parsing is linear in the page size, so no page can stall the run. Sends the apify-actor-fleet/1.0 User-Agent.
• A response in an encoding this Actor cannot decode (for example zstd) is not parsed: you get a free UNSUPPORTED_ENCODING row instead of a metadata row full of false issues.
• No JavaScript rendering, no crawling, no link checking. It reports what the server sends.
• A site behind a bot-protection wall may answer 403; that is returned as a free HTTP_ERROR row with the real status.