CollimerBot

If you found CollimerBot in your server logs and followed the link, this page is for you. It explains what the crawler is, why it visited, and how to stop it.

What CollimerBot is

Collimer audits how a website appears in AI search assistants such as ChatGPT, Claude, Gemini and Perplexity. Part of that audit reads publicly accessible pages of the site being evaluated, so the report can say what those pages actually contain rather than guessing.

CollimerBot is the crawler that does the reading. It requests pages the way a browser would. It does not submit forms, does not attempt to log in, and does not follow links into anything behind authentication.

Why it visited your site

Someone ran an audit that named your domain. That happens in one of two ways:

1. They audited their own site, and it is yours.
2. They audited a competitor set, and your site was named in it.

The second case is why you may be seeing us without having asked for anything. If you would rather we did not fetch your pages, the next two sections are the answer.

How to identify it

CollimerBot sends this user agent on every request:

Mozilla/5.0 (compatible; CollimerBot/1.0; +https://collimer.com/bot)

Requests come from Fly.io infrastructure. We do not publish a fixed IP range, so match on the user agent string rather than on addresses.

How to stop it

Today, the way that works is email. Write to hello@sandcastlelabs.ai from an address at the domain, or tell us how you are connected to it, and ask for the domain to be suppressed. We do it by hand and confirm when it is done.

Suppression does two things:

• We stop fetching the domain’s pages.
• Any public scorecard page for that domain stops being served, including ones already generated.

robots.txt support is being built and is not live yet. We would rather say that than print a stanza that quietly does nothing. When it ships, this is what will work, and this page will be updated with the date it landed:

User-agent: CollimerBot
Disallow: /

If you add that now, it starts being honoured the day support lands, and it costs you nothing in the meantime. Standard robots.txt syntax applies, so you can disallow specific paths instead of the whole site.

What suppression does not cover

Collimer’s core measurement asks AI assistants questions and records how they answer. That measurement never touches your servers, and it reflects what a model says rather than anything we publish about you. We cannot stop a model from naming your brand, and suppression does not attempt to.

We are telling you this plainly rather than letting the word “suppression” imply more than it delivers. If a model describes your company inaccurately, the fix is on the surfaces the model reads, not with us.

What we store

Pages fetched by CollimerBot are stored as part of the audit that requested them, under the retention described in our Privacy Policy. Suppression removes the public scorecard for a domain. It does not retroactively delete audits already run inside a customer’s private account.

Questions

Email hello@sandcastlelabs.ai. A person reads it.