For site owners
TragentBot
TragentBot is the crawler behind tragent, a service that checks whether online stores work for AI shopping agents. If you found it in your logs, this page explains what it did and how to stop it.
User agent
TragentBot/0.1 (+https://tragent.io/bot; agent-compatibility-check)
The robots.txt token is TragentBot.
Why it visited
- A public scan. Someone entered your site in the tragent scanner. A scan reads a handful of public pages and files (such as the home page, a product, robots.txt and structured data), honours robots.txt, and is rate-limited per domain.
- A quick store check, AI shopping test or scheduled check. A tragent customer is testing a site they operate. AI shopping tests and scheduled checks only run against domains the customer has proven they control with a DNS record or a file, so this traffic was requested by the site owner (or someone they authorised).
Other user agents in the same runs
Part of every scan and quick store check asks whether your bot management lets real AI agents through. To answer it, tragent requests a few pages presenting the published user-agent strings of well-known agents (for example ChatGPT-User or Claude-User), with and without signed-agent headers, and records whether they are blocked. Those requests come from tragent's infrastructure, not from those companies, and are part of the same small, rate-limited visit. Simulated agents in AI shopping tests may likewise use the user agent of the agent style being tested.
Limit or block it
Public scans honour robots.txt. To stop them entirely:
User-agent: TragentBot Disallow: /
Runs a site's own team schedules are under their control; if that is you, pause the scheduled check in the dashboard. If you do not recognise the traffic, or want your domain excluded from public scans, email [email protected] with the domain and a few log lines. See also our scanning policy.