What this shows A real headless Chrome runs at the edge from a Worker on the Workers Paid plan. Paste a URL. A Workflow saves a desktop screenshot, a 390px phone screenshot, a PDF, and markdown, then Workers AI writes a short summary.
A cron deletes capture rows and their R2 files about a day after they are created. It runs every 15 minutes, so a file can remain a little longer. An R2 lifecycle rule is the backstop. No account.
What you can do
Paste a URL, pass Turnstile, and watch each workflow step.
Scroll the desktop and phone screenshots inside the frame.
Open the PDF, read the markdown, and read the summary.
See whether the capture reused the warm browser or launched a new one.
Try a blocked address, such as loopback or a metadata host, and read the rejection.
Capture a page
6 captures per browser and per IPv6 /64 each hour. 20 per IPv6 /48. 300 a day for the demo.
Recent captures
This browser only. Opening one needs its unguessable id.