The capture browser
What visits a site for SuperPost, the names it sends, and how to allow or block it.
What it is
SuperPost shows a product as it really is. To do that, its capture service opens the product’s public web address in a browser, takes screenshots, records a few seconds of video, and reads the page’s title and description. A person using SuperPost names the address of their product. The Pull Test also opens the homepage listed on the public GitHub repository it reads, once.
What it does on a site
- It opens pages, scrolls and moves the pointer over links. It never clicks, types, fills in a form or signs in.
- The browser loads only the host it was given; requests to any other host are not resolved.
- It uses HTTPS on the default port only, and refuses private and internal addresses.
- It never tries to get past a CAPTCHA, a second sign-in step or a firewall, and it does not disguise itself. If a site blocks it, the capture stops.
The names it sends
- When the capture service reads a page’s HTML, it sends the user-agent
SuperPost-Capture/1.0. - When SuperPost’s own server reads a page, it sends
SuperPost/1.0. - The browser that takes screenshots and recordings does not carry SuperPost’s name yet. It sends the standard user-agent of headless Chromium, which contains
HeadlessChrome.
Its address
There is no fixed outbound address yet: the capture service runs on one machine whose address is not reserved and can change. This page will give one fixed address you can allow before SuperPost starts capturing other people’s sites.
How to block or allow it
- To block page reads, refuse the user-agents
SuperPost-Capture/1.0andSuperPost/1.0. - To block the browser today, block headless Chromium. That also blocks other tools that use it.
- To let page reads through a firewall or bot protection, allow those two user-agents. The browser has no name or address of its own to allow yet.
- It does not read robots.txt.
Questions about a visit
Write to security@superpost.io with the time and the address it visited.