MintyOrangeBot.

MintyOrangeBot is the page fetcher behind Minty Orange, a content optimization tool for Google and AI search. It fetches pages in two situations, and in both a person inside our product asked for it. In almost every case that person owns the site.

What it fetches

Single pages. When someone asks us to analyze or optimize one page, we download that page once. Nothing else on the site is touched.

Site audits. When a site owner asks for an AI visibility audit of her own site, we read her sitemap and download the pages listed there: up to 1,500 pages for scoring, and up to 2,000 for the internal link map. This runs at one request every 2.5 seconds, never in parallel, and at most once a month per site. A page that fails to load gets one more try at the end of the audit, at the same pace. Site owners start audits themselves; we do not audit sites nobody asked us to.

What it does not do

It does not follow links beyond the sitemap, it does not build an index, and it does not collect training data. Outside of audits the whole service makes a handful of requests per day. Every request maps to one human action inside our product.

Until September 2026 this page said the bot does not crawl. That was true for single pages and wrong for audits, which back then also ran with several requests in parallel. We corrected both.

How it identifies itself

Requests use this User-Agent:

MintyOrangeBot/1.0 (+https://mintyorange.ai/bot)

If a site rejects the declared bot, we may retry the same single page with a standard browser User-Agent. We would rather be transparent about that fallback than pretend it does not exist. It exists because many bot protections block any unknown bot name by default, even when the page owner is the one asking for the fetch.

How to verify it is really us

Two independent ways, and you do not have to take our word for either.

Reverse DNS. Our requests come from a fixed address whose reverse lookup resolves to crawler.mintyorange.ai, and that name resolves back to the same address. This is the same check used to verify Googlebot.

Signed requests. Every request carries an Ed25519 signature over the target host, as described in RFC 9421 and the Web Bot Auth drafts. Our public key lives at /.well-known/http-message-signatures-directory. A forged request cannot produce a valid signature, so this holds even if someone copies our User-Agent.

How to block it

We honor robots.txt rules addressed to MintyOrangeBot by name. To keep it away from your whole site, add:

User-agent: MintyOrangeBot
Disallow: /

Because every request is triggered by a person working on one specific page, we do not apply generic User-agent: * rules, the same way other user-triggered fetchers work. A rule that names MintyOrangeBot always wins, and a named Disallow stops the fetch before it happens.

Contact

Questions, complaints, or a block you want lifted or added: write to hello@mintyorange.ai or use our contact form. We answer fast. There are real people here, and not many of us.