Teardown judges an online shop's design for the person who runs it. TeardownBot is how it reads the shop - and it only ever reads the pages a shop's owner has asked about. It does not crawl the web, follow links to other sites, sign in, add anything to a basket or go through a checkout.
The reader fetches a page's HTML, its stylesheets and the first few bytes of its images, to check them against fixed rules. It sends this user agent:
TeardownBot/1.0 (+https://www.teardown.space/bot; reads shop layout for its owner)
It reads robots.txt first and does what it says, including a Crawl-delay.
The browser opens the same pages in Chromium to take one picture of each, as it looks on a desktop and on a phone, for the AI review of how the page looks and reads. Its user agent is an ordinary Chrome one with this added to the end:
TeardownBot/1.0 (+https://www.teardown.space/bot; judges shop design for its owner)
It only opens pages the reader was allowed to read. While it does, it skips analytics and advertising scripts, and never plays video.
Both run on our host's servers, so there is no fixed list of addresses to allow. Allow the name TeardownBot in the user agent instead.
Every request is also signed (HTTP Message Signatures, as Cloudflare's Web Bot Auth uses them), so a firewall can check it really is ours and not a copy of our name. Our public key is at https://www.teardown.space/.well-known/http-message-signatures-directory.
Only if you want to - a report still comes back without it, and you can upload your own screenshots of any page it could not see. If you do:
TeardownBot, with the action Skip. Cloudflare's free
Bot Fight Mode cannot be skipped by a rule; switch it off while your report runs, and
back on afterwards.TeardownBot to be
allowed.Then open your report and press I have let it in - try again.
Add this to your shop's robots.txt. The reader will stop, and the browser only goes where the reader may:
User-agent: TeardownBot Disallow: /
Write to mainproductioncl@gmail.com.