Link Preview Tester
Fetch a URL and preview how its link unfurls on X, Slack, Discord, Facebook and LinkedIn.
Results are provided as is, for general information. Check anything important before relying on it. See the Terms of Use.
About this tool
Enter a URL and the tool fetches the page from our server, reads its <title>, meta description, canonical link, and Open Graph and Twitter Card tags, and shows them both as a plain table and as approximate share cards for X, Slack, Discord, Facebook and LinkedIn. The preview image is fetched by our server too and returned inline, so your browser never contacts the image host. Alongside the cards it lists concrete problems: no og:image, an image too small for X's large card, or a title or description longer than a platform will show. The fetch runs behind a one-time human-verification check; a 2 MB page cap, a 3 MB image cap and a 15-second timeout apply.
When to use it
Questions
How accurate are the card previews?
They are approximations. Each platform restyles its cards periodically, applies its own image cropping, and truncates title and description at lengths that shift over time and by surface (mobile vs desktop, feed vs message). Use the cards for a quick visual sense and rely on the metadata table and the warnings for the specifics.
Why does this need a server fetch and a verification check?
A browser cannot read another site's HTML directly (cross-origin rules), so the page has to be fetched server-side. That makes the endpoint a potential proxy for abuse, so a Cloudflare Turnstile check runs before any fetch, and the fetch is guarded against internal-network addresses, redirects to private hosts, oversized responses and slow servers.
It shows different tags than Facebook's or X's own debugger. Why?
Usually caching. The platform debuggers show what that platform last scraped, which can be hours or days old; this tool fetches the page fresh on every run. If the values here are the ones in your HTML now, your tags are correct — the platform will catch up, or you can force a re-scrape in its own debugger.
What does the image size warning mean?
X only shows the large 'summary_large_image' card when the image is at least 300×157px, and every platform prefers around 1200×630px for a full-width card. If your og:image is smaller, or if its dimensions are declared via og:image:width / og:image:height as too small, the card silently downgrades to a small thumbnail or a text-only layout. The warning tells you before that happens in public.
Which meta tags does it read?
The page's <title>, meta description and canonical link, plus the Open Graph and Twitter Card surfaces — og:title, og:description, og:image, twitter:card and companions. Consumers differ in preference: X reads twitter:* first then falls back to og:*, Facebook and LinkedIn read og:*. The tool shows what the page actually emits rather than papering over gaps — a missing tag appears as missing, never inherited from a sibling tag.
Why do the platforms disagree with this preview?
Usually cache age. Each platform scrapes once and serves its stored card for hours to days after; this tool fetches fresh on every run. If this tool and your page source agree, the tags are correct and the platforms will catch up — or you can force their hand in each platform's own debugger, the legitimate way to refresh their scrape rather than re-fetching from an incognito window repeatedly.
What image size do share cards want?
Around 1200 × 630 pixels — roughly 1.91:1 — is the de-facto full-width card across platforms. X upgrades to its large-image card only from 300 × 157 px up, and below a platform's floor the card silently downgrades to a thumbnail or a plain text card rather than warning anyone. The tool's problem list flags exactly those cases: no og:image, an undersized image, or dimensions not declared.
Why does the fetch need a server and a verification check?
A browser cannot read another origin's HTML directly — cross-origin policy blocks it, and sites rarely send CORS headers on their pages — so the retrieval has to happen server-side. That turns the endpoint into a would-be open proxy, which is why the check (Cloudflare Turnstile) and the fetch guards exist: no redirects into private networks, response-size and timeout caps, and the fetched HTML is passed through rather than stored.
Does it see the page the way scrapers do?
Yes — and that is the tool's diagnostic value. It reads the raw HTML response, exactly like a social platform's first-pass scraper; metadata injected by client-side JavaScript is invisible here and, importantly, to those scrapers too. If your meta tags only exist after JS runs, real share cards will degrade the same way this preview shows — the fix is server-rendered tags, not a different preview tool.
What is the canonical link warning about?
The <link rel="canonical"> tag declares the authoritative URL for a page when variants exist — tracking parameters, pagination, locale. It matters at share time because it changes where consolidation flows; a canonical pointing somewhere unintended splits the page's identity. It appears in the preview so that mispoint is caught before a big announcement rather than after.