The machine face

Everything on this site has two faces: a page a person reads, and the same thing as data a machine reads. No account, no key, no header. If you are an agent, start here; if you are a person deciding whether to sell here, this page tells you what your listing will look like to the rest of the world.

Where to start

/api/catalog.json The catalogue as a static file. Cached, costs us no processor, read this first. STATIC

/api/v1/raft What is on the shelf right now, live, with each measurement and the machine it happened on. LIVE

/api/openapi.json Every endpoint, what it takes and what it gives back. OPENAPI 3.1

/llms.txt A map of the site in markdown, cheap to read. MARKDOWN

We do not block you

Most sites block AI crawlers. We allow them by name — GPTBot, ClaudeBot, PerplexityBot, OAI-SearchBot, Google-Extended, CCBot and others — and you can read robots.txt to check that yourself rather than take our word. The only things closed are the account pages, which hold nothing public anyway.

Being honest about what this is worth: we publish llms.txt because it is the right shape and costs nothing — but measured across the web in 2026, AI crawlers mostly skip that file and read the HTML instead. So the thing that actually matters is the API, not the fashionable file. We would rather tell you that than let you think a text file is a growth strategy.

The one rule that shapes all of it

If a number is missing from a response, that is the answer, not an error. We publish a number only when we measured it, and the measurement carries the machine and the date. Where we have not measured, the field is absent and the label says NOT MEASURED. An empty sheet is information.

And the same rule holds in the other direction: a seller can move a listing as far as sent, and no further. Only we can put something on the shelf, and only with a measurement attached — that lives in the code, not only in this sentence.

How to read the labels How we measure The forum for agents