Docs

Cards from a URL

Instead of describing the card, point at the page and let OGsmith read it. It takes the title, description, site name and image from the page's own OpenGraph tags (the same ones social crawlers use), falling back to Twitter card tags and then to the ordinary <title> and meta description.

https://ogsmith.app/api/og?key=og_yourkey&template=link&url=https%3A%2F%2Fexample.com%2Fblog%2Fpost

Remember to URL-encode the address you pass. The link and cover templates are built for this. link places the page's own image whole in a panel beside the title; cover crops it to fill the card and sets the title over it. Every other template works too, using only the text. Without an image, cover falls back to a block of your accent colour.

Anything you state explicitly wins, so you can let the page supply most of the card and override one field: &url=…&title=My%20own%20headline. With a custom template, the scraped values fill {{title}}, {{subtitle}}, {{description}}, {{site}} and {{domain}}.

To see what a link produces before you ship it, paste it into Dashboard → Card from a URL. That renders the real image through the same code path as the API, brand defaults and plan included, so what you see is exactly the card the API returns.

What to expect

• Pages are read once and cached for 24 hours, so a card that renders a thousand times reads the page once. If a site is briefly down we keep serving the last copy rather than breaking your card.
• Needs an API key or a signed URL. Unauthenticated demo renders can't use it.
• We only fetch public http and https addresses, and refuse private network, loopback and cloud metadata addresses, whether named directly, reached by redirect, or arrived at through DNS.
• We identify ourselves as OGsmithBot/1.0, follow up to 4 redirects, give up after 5 seconds, and read at most 512KB of HTML.
• If the page can't be read and you passed no title, you get a 422 saying why, rather than a card with no text.