Cloudflare Browser Run adds PDF, links and JSON to one snapshot
The /snapshot endpoint can now return four more formats from a single page load, cutting the round trips needed to scrape a page or pull structured data from it.
Cloudflare published a changelog update today extending Browser Run’s /snapshot endpoint to four new output formats: PDF, links, scraped elements and AI-extracted JSON. Before this, /snapshot only returned HTML, a screenshot, markdown and an accessibility tree from a page load.
Each new format reuses the options from its dedicated endpoint. PDF takes the same pdfOptions as the /pdf endpoint, links reuses visibleLinksOnly and excludeExternalLinks, scraped elements needs an elements parameter, and JSON needs a prompt or response_format and runs through Workers AI, so it bills separately. You still need to request at least two distinct formats, and they’re captured in a fixed order with PDF printed last. If any one capture fails, Cloudflare fails the whole request, since the page can change between captures. For JSON requests, a single actionTimeout budget (60 seconds by default, 120 at most) covers everything.
{
"url": "https://example.com",
"formats": ["pdf", "links", "scrape", "json"],
"elements": [{ "selector": "p" }],
"prompt": "Summarise the page"
}
Why it matters: anyone scraping a page or feeding it to an agent can now pull a PDF, a link list, matched elements and structured JSON from one render instead of four separate requests, which saves both render time and round trips.