MetaPreviewMETADATA PREVIEW

SERP · Open Graph · X card · JSON-LD · one page

See the result before you publish it

One title, one description, one image, and you can see the Google result, the Facebook, KakaoTalk, LinkedIn and Slack cards and the X card side by side — with the length that actually gets cut marked in pixels, not guessed. The same fields produce the meta tag block, the JSON-LD, a GEO checklist and an llms.txt snippet you can copy or download. No account, no upload, no call to any server.

Surfaces Google · Facebook · KakaoTalk · LinkedIn · Slack · XStructured data 4 schema types, validatedGuidance checked 21 November 2026Network 0 requests from the tool

Metadata preview tool

Page details

A chosen file is read in this tab only. A remote URL is never fetched, so its size stays unverified.

    Drawn at Google's published sizes; the favicon slot shows the site's initial because this tool fetches nothing. Google sometimes rewrites a title or description from page content, which the preview cannot predict.

    GEO / SEO checklist

    Eleven independent checks on the values above. A warn is a broken guide, not a broken page.

      llms.txt snippet

      A Markdown summary for answer engines. Save it at the site root as llms.txt.

      
          

      Features

      What the page shows you

      Six things, all without a server.

      SERP

      Google result, desktop and mobile

      The favicon, site name, breadcrumb, blue title and two-line description are drawn at Google's own sizes: 20px title in a 600px line on desktop, 18px on mobile. The title and description are measured with the browser's text metrics, so you see where the cut really falls, not just a character count.

      OG

      Four social cards from one image

      Facebook, KakaoTalk, LinkedIn and Slack each draw the same og:title, og:description and og:image in their own layout, so a card that works on one does not quietly break on another. A local image is read in the tab and shown at its real proportions against the 1.91:1 guide.

      X

      summary_large_image or summary

      Switch the card type and the X preview changes with it, including what the small square image does to the layout. The twitter tags in the code block follow the same choice.

      JSON-LD

      Four schema types, checked

      WebApplication, SoftwareApplication with price 0, Article and FAQPage are built from the same fields, then checked for syntax and for the properties each type requires. A pasted block can be validated the same way before you ship it.

      GEO

      Canonical, hreflang, robots, llms.txt

      Eleven checks cover the tags answer engines and crawlers read: absolute URL, canonical, the four-language hreflang set with x-default, robots, Open Graph, image ratio, X card, JSON-LD and an llms.txt snippet generated from your fields.

      LOCAL

      Nothing leaves the tab

      No fetch, no XHR, no beacon. The tool is one script; your fields stay in this browser's localStorage and an uploaded image is read with the File API. Clear site data and both are gone.

      How it works

      Four steps, no setup

      The previews redraw on every keystroke; there is no run button.

      01

      Type the page details

      URL, site name, title, description and an image. Load the sample to see a filled-in example, or start blank.

      02

      Compare the previews

      Google at desktop and mobile width, the four social cards and the X card. Length meters say what will be cut and by how much.

      03

      Pick a schema type

      Choose WebApplication, SoftwareApplication, Article or FAQPage; the required fields are filled from your page details.

      04

      Copy or download

      The meta tag block as HTML, the JSON-LD as .json, the llms.txt snippet as .txt — each with one button.

      Coverage

      What each surface reads

      Six previews, one set of fields. The last column is where each one stops.

      SurfaceReadsWhat the preview checksWhere it stops
      Googletitle, meta description, URL path, favicon, site nameTitle cut at 60 characters or 600px, description at 160 characters, breadcrumb from the URL pathGoogle rewrites some titles and descriptions from page content; the preview cannot predict that.
      Facebookog:title, og:description, og:image, og:urlImage at 1.91:1, title and description as the feed shows them, domain lineFacebook caches a card once it is shared; changes appear after its crawler refetches.
      KakaoTalkog:title, og:description, og:imageLarge image with the title and description under it, as the chat bubble draws themThe exact crop depends on the KakaoTalk version and the device width.
      LinkedInog:title, og:description, og:imageHeadline, domain and description under the imageLinkedIn sometimes shows a smaller thumbnail layout for short posts.
      Slackog:title, og:description, og:image, og:site_nameUnfurl block with the site name, title and description beside the imageSlack shortens descriptions and hides the image for some domains.
      Xtwitter:card, twitter:title, twitter:description, twitter:image (falling back to Open Graph)summary_large_image and summary layouts, image proportions, character metersX crops a large image to about 2:1 in the timeline and shows no card for some links.
      JSON-LDyour fields plus the schema type you pickSyntax, @context, @type and the required properties per typeValid markup does not guarantee a rich result; the search engine decides.
      llms.txtsite name, URL, title, descriptionA Markdown summary in the shape language models readllms.txt is a convention, not a standard; not every crawler reads it yet.

      Every limit here is a guide, not a rule: search engines and chat apps change their cuts without notice. The meters say which number was used so you can judge the margin yourself.

      Under the hood

      One field set, measured once, drawn six ways

      The previews are not screenshots: each one is HTML drawn from the same values, measured in the browser.

      your fields→normalise→measure→draw→previews · code · llms.txt
      Normalise trim, collapse whitespace, resolve the URL and its pathMeasure canvas text metrics at Google's sizes, image pixels from the File APIDraw one render function per surface, text inserted as text, never as markup

      The Google preview measures text instead of counting it. The browser lays the title out at 20px Arial and the tool reads where the line passes 600px, which is why a wide character can be cut earlier than the sixtieth character. The character count sits beside the pixel count so both readings are visible.

      Images are treated carefully. An uploaded file is read locally and drawn from a blob URL, so its true width, height and ratio are known. A remote og:image URL is never fetched — that would be a network request — so the card shows a placeholder and the checklist reports the ratio as unverified rather than guessing it.

      The JSON-LD is assembled from a plain object and serialised with JSON.stringify, then parsed again for the check, so what you copy is exactly what was validated. The checklist re-runs on every keystroke with it.

      Questions

      What people ask before using it

      Does MetaPreview upload my page or my image?

      No. The tool is a single JavaScript file that runs in the tab. The fields you type are kept in this browser's localStorage, an uploaded image is read with the File API and shown from a blob URL, and no fetch, XHR, WebSocket or beacon request is made. The only external request on the page is the AdSense loader in the head, which the site owner added for ads.

      Why is the preview only an approximation?

      Because the real result is decided by each platform. Google chooses the font, the pixel width and sometimes rewrites the title from page content; Facebook caches a card after the first share; X crops a large image to about 2:1. The preview reproduces the current layout and the published limits, and the meters say which number was used, so you can leave a margin rather than trusting one exact pixel.

      What does the 1.91:1 image guide mean?

      It is the shape Open Graph cards prefer: 1200×630 pixels, or any image whose width is about 1.91 times its height. An image far from that ratio is cropped by the platform — usually the top and bottom. Upload your image and the tool measures the file itself and says how far the ratio is from the guide.

      Do I need a JSON-LD block to rank?

      No. Structured data does not raise a ranking by itself; it makes the page machine-readable so a search engine can show a richer result when it chooses. The generator covers the four types most sites need — WebApplication, SoftwareApplication, Article and FAQPage — and validates syntax and required fields, which is where most broken markup fails.

      What is llms.txt and do I need one?

      llms.txt is a proposed convention: a Markdown file at the site root that summarises the site and its main pages for language models and answer engines. It is not a standard and no crawler is obliged to read it. The generator writes a starting file from your fields; treat it as a tidy summary, not a ranking lever.

      Is the checklist a score?

      No. It lists eleven separate things — title length, description length, absolute URL, canonical, hreflang set, robots, Open Graph tags, image ratio, X card tags, JSON-LD and llms.txt — each marked pass, warn or fail on its own. A warn means the value breaks a published guide, not that the page is bad.

      Limits

      What this page does not claim

      Previews are drawn, not screenshotted

      Each card is HTML built to match the platform's current layout. Fonts, spacing and crops can differ by a pixel or two, and a platform can redesign without notice.

      Pixel limits are the published guides

      600px for a desktop title, 920px for a description, 1.91:1 for an image: these are the numbers the meters use, checked on 21 November 2026. They are guides, not guarantees.

      A remote image is never fetched

      If you give an og:image URL rather than a file, the tool draws a placeholder and reports the ratio as unverified. Fetching it would be a network request, and this tool makes none.

      Valid JSON-LD is not a rich result

      The generator checks syntax and required properties. Whether a search engine shows a rich result, and what it shows, is the search engine's decision.

      Your fields stay in this browser

      They are stored in localStorage on this domain so a reload does not lose them. Nothing is sent anywhere, and clearing site data removes them.

      Where the numbers come from

      • Title — about 60 characters or 600 pixels on desktop, 18px type on mobile
      • Description — about 160 characters, roughly 920 pixels in the desktop result
      • Open Graph image — 1200×630 recommended, minimum 200×200, ratio about 1.91:1
      • X card — summary_large_image: 2:1 crop in the timeline; summary: square thumbnail
      • JSON-LD — schema.org types WebApplication, SoftwareApplication, Article, FAQPage
      • llms.txt — a proposed Markdown convention at the site root, not a standard