Skip to main content

xbrushbot-link

xbrushbot-link/1.0 (+https://xbrush.ai/docs/bot; link preview) is the user agent XBRUSH uses to build a link preview.

When it visits a page​

It visits a page only when an XBRUSH user pastes that page's link into their workspace chat.

  • One request per pasted link, at the moment of pasting.
  • If the user then asks the AI to use the link, the page text or the preview image may be fetched once more, at that moment.
  • No crawling, no link discovery, no scheduled visits. Every request is made on behalf of a person for a single link they chose.

What it reads​

  • Title, description and preview image URL (Open Graph, Twitter Card or <title>).
  • On request, the main text of the linked page.
  • YouTube and TikTok are read only through their official oEmbed endpoints.

Limits: 5 seconds, 2 MB, text/html only, HTTPS only, up to 3 links per message, rate limited per user.

What it keeps​

Only the title, description and preview image URL, as part of the user's message. Page text and images are not stored.

robots.txt and AI opt-out signals​

Because each request is a single fetch made for a person who chose the link, xbrushbot-link works like other user-initiated fetchers (such as ChatGPT-User or Claude-User). It does not apply crawler rules from robots.txt, and it does not treat AI-training opt-out signals (such as noai or tdm-reservation) as covering this use: nothing it reads is used for training, and page content is not stored.

How to block it​

Block the user agent xbrushbot-link/1.0 at the network or application level. For questions about a specific request, contact us below.

Contact​

support@xbrush.ai — for personal-data matters, use the contact listed in our Privacy Policy.

Last updated on