HTML metadata inspect
Inspect an HTML document and download a structured metadata report: title, description, canonical URL, robots, Open Graph, Twitter/X card tags, language, charset and more.
Working…
Questions people ask
- Is this an SEO scoring tool?
- No. It is an offline document inspector: it reports the metadata a document declares, verbatim. There is no scoring, no advice and no crawling: the tool never reads a second document, fetches a URL or executes anything.
- What is reported, exactly?
- The document's own declarations: the title element, name-based meta tags (description, robots, author, generator, viewport, theme-color and friends), Open Graph and Twitter/X card tags (both the property= and name= spellings), the canonical link, the icon/apple-touch/alternate/manifest/amphtml link family, charset and http-equiv declarations, html lang and dir, and the base href. A declaration is included only when present in the document.
- How do you tell absent from empty?
- Strictly: a key the document never declares does not appear in the report at all (absent), while a declared-empty value appears as "" (empty). The UI summary likewise reads "declared" or "absent", never a guess that stands in for a missing tag.
- What happens with duplicate or conflicting tags?
- Every declaration is listed, in source order, with the count; the summary value is the first occurrence verbatim. The same tag spelled two ways (og:title via property= and name=) is reported as two declarations. Conflicts stay visible and are never silently resolved.
- Is a relative canonical resolved against the base tag or some origin?
- No. Values are reported exactly as declared. If the document carries a <base href>, that is its own reported fact. The tool never combines it into your canonical or other URLs, because the resolution would be inferred, not declared.
- Is parsing safe? Does anything execute or get fetched?
- No execution and no fetching, ever: the document goes through a non-executing parser. Hostile hrefs are just reported text; malformed markup is recovered the same way the HTML-to-Markdown tool recovers it. NUL-byte or bad-UTF-8 documents are refused with named diagnostics.
- How big can the document be?
- Up to 4 MB of HTML; the report is capped at 2 MB (a report can only grow near the size of its own declarations), and a per-key declaration list beyond five entries is flagged rather than truncated silently.
- What happens to my files?
- The input is removed as soon as processing finishes; the report is available for about 10 minutes, then automatically deleted.