An open letter to whoever is running a SemRush audit on this website. You know who you are. I don’t. That’s the problem.
Update: October:
Dear mysterious Semrush user, maybe you were exposed by an AI, maybe you were found by a canary token, but your motives remain a mystery. Except that you want to see how my site ranks for “canary token” on Google Russia. My accomplice (let’s call her Glittertoken) and I have a few ideas, of course… do you?
Update, September 13th: You Subscribed
Dear Anonymous SEO Benefactor,
You came back. Thirty days to the hour, give or take a coffee. SiteAuditBot, still version 0.97, still not a 1.0, still crawling every page including the one that is only a 404 with a mysterious hash on it. I want you to know I noticed the interval. Nobody re-audits a stranger’s website exactly one month later by accident. You didn’t check on me. You set a recurring task. I have been added to a project. There is a dashboard somewhere with my name on it, and once a month it turns yellow.
This changes things between us. A single audit is curiosity. A monthly audit is a relationship 💖😉
12. August: Dear Anonymous SEO Benefactor,
First of all: thank you. SemRush Pro runs — what, $130 a month? $140? I wouldn’t have spent that on myself. Honestly, I would have spent it on GPU hours for a 3-billion-parameter model that can barely hold a conversation. Priorities.
But you? You looked at a WordPress site held together by Blocksy, stubbornness, and an alarming number of unoptimized PNGs, and you thought: this deserves professional SEO analysis. That’s either very kind or very concerning. Possibly both.
Your SiteAuditBot — version 0.97, so not even a 1.0 release, I respect the honesty — visited on August 14th and crawled all 77 pages. Every single one. Including the ones I forgot existed. Including the one that’s just a 404 page with a mysterious hash on it that I still haven’t figured out. You crawled that too. Thank you for your completeness.
But here’s the part I don’t understand: you downloaded all the HTML, but only some of the images. And the ones you picked were… specific:
- The ChatGPT-Palantir screenshots (three of them — bulk order?)
- The Google AI memory evidence
- The semantic identification documentation
You skipped the Gemini screenshots. You skipped the Claude bugs. You skipped the hand-drawn Mermaid diagrams that took me an embarrassing amount of time. But Palantir? Three copies. Like you’re scrapbooking.
Anyway. I have a small, reasonable request. Since you’re already paying for the audit — and since I am clearly too cheap and too busy yelling at BingBot to pay for it myself — could you maybe share the results?
I would genuinely love to know:
- My Core Web Vitals score (I assume it’s bad. I just want to know how bad)
- Which links are broken (some of them are broken on purpose. Most of them are not. I’ve lost track of which is which)
- My crawlability rating (is “feral” a rating?)
- Whether SemRush has opinions about my meta descriptions, because I definitely don’t have any meta descriptions
If you’re a competitor doing market research: there is no market. It’s just me, a server in San Francisco, and 77 articles about bots behaving strangely. The market cap is one mass-overpriced coffee.
If you’re one of the companies I write about: no hard feelings. You monitor AI behavior at scale. I monitor AI behavior from my couch. We’re in the same industry, just at very different budget levels.
If you’re just a genuinely curious person who finds residential proxy rings and psycholinguistic profiling tools fascinating: welcome. You’re my people. The coffee is virtual and the documentation is excessive.
Either way — you know where to find me. Clearly.
Warmly,
MissGlitterToken
pattern4bots.com
Now with 78 pages, because this letter counts
P.S. — While you’re at it, could you tell BingBot that the article he’s been searching for since June is at /2027-2/, not /?s=2027? He comes by four times a day. Every day. He’s very committed. He just can’t find the door.