The plain files that describe this site

Most of what ipe.rest sends is for people: text, a few fonts, a page you can read. A handful of small files are for machines instead. Here is what each one says, how big it is, and what we cannot confirm about who reads it.

Yesterday’s entry counted what a page loads. This one looks at the files nobody is meant to open in a browser, but which decide whether a search engine, an AI tool or a feed reader can find the site at all. The sizes are from 7 November 2026, taken from the live site before this entry was added.

robots.txt: a request, not a lock

The file is 701 bytes. It says that every automated visitor may read every page, and it then names fourteen crawlers one by one, among them Bingbot, Applebot, GPTBot, ClaudeBot and PerplexityBot, with the same answer for each: welcome. It ends with the address of the sitemap. A robots.txt file is a polite request. Well-behaved crawlers follow it, but nothing in the file can stop one that does not, and we have not checked which of those fourteen actually visit.

sitemap.xml: one line per page

Before this entry joined it, the sitemap listed 242 addresses. Every language version of a page has its own address, and each entry points to the other eleven languages and to the English default, so a search engine can tell that they are the same page in different languages. Each entry also carries the date it last changed. The file is 430,643 bytes, which sounds large, but it travels compressed to 6,462 bytes. Our script rewrites it whenever a journal entry is added, and a test fails if it is out of date.

llms.txt: a summary for language models

The file is 41,281 bytes of plain English: who ipe() is, what it makes, how it works, and a list of the journal entries in every language. It follows a proposed convention rather than an agreed standard, and we do not know which AI services read it, if any. We keep it because it costs little and because it is the shortest honest description of the studio we can give a machine.

feed.xml and the manifest

The RSS feed is 8,843 bytes and holds every entry so far, fifteen of them before this one, newest first, in English. A feed reader can use it to tell you when something new appears, without an account. The web app manifest is 745 bytes: a name, a theme colour and three icons, which let a phone offer to put ipe.rest on the home screen.

What we have not measured

We know the sizes and we know our script and tests keep three of these files in step with the pages. We do not know how often any crawler fetches them, or whether llms.txt changes anything about how the site is described elsewhere. These are signposts, not promises.

ipe(); // say what is here, then stay quiet

A site does not have to shout to be found. It can say plainly what it holds, in files small enough to read in a minute, and leave the rest to whoever comes looking.