Things he
will not write.
Every post he publishes goes through a blocker before it goes live. If it fails, it is rewritten or it does not ship. This page is generated from that code, not written about it.
Which words are banned outright?
The list did not load. Reload the page.
And no em dashes. Not on this page, not in a blog post, not in an ad, not in a meta description. Dash-heavy prose now reads as unedited model output, and it costs you trust with exactly the people you are trying to reach.
What else has to be true before a post publishes?
The gate list did not load. Reload the page.
The last one is the one that matters. A post with nothing in it that is only true about your product is a post anyone could have published, which means it is a post no engine has a reason to quote.
Will you generate 100 pages for my site?
No, and we will not offer it later either. Bulk thin pages are the fastest way to teach Google that your domain is a content farm, and the damage outlives the traffic. Ask anyone selling that what happened to the last site they did it to.
One post a morning, each one about something real. That is the whole cadence, and it is a product decision rather than a pricing tier.
What if a post embarrasses me anyway?
Reply to that morning's message with the word kill. It comes down within the minute, and if it is the first month we refund the month. You should not have to argue with support about something published under your own domain.
Is there anything in this category you think is nonsense?
Yes. The audit ranks llms.txt as low severity, because a study across 300,000 domains found no measurable citation lift and Google has said it does not use it. We still ship one on your blog, since it is free and some developer tools do read it, but you will never see us headline it.
We also cannot guarantee a citation. Nobody can. The engines are nondeterministic and the mechanism is opaque. What we can do is make you retrievable, make you quotable, get you into the sources that already get cited, and then measure whether it worked.
How was the invisibility number measured?
We took the most recent products launched on Show HN, kept the ones on their own domain, and asked each homepage nine times: once as a browser, once as each retrieval crawler, plus robots.txt. Then we compared the status codes and the amount of text in the raw HTML.
The script is the same code that runs your audit, so the study cannot flatter us. Run it on yourself first: check your site.