The endpoints
Everything is a static file. No authentication, no API key, no pagination.
Each wall page declares its markdown twin with a rel="alternate" link, so a crawler that lands on the HTML can find the clean version. The fifteen wall slugs are listed in llms.txt and in the JSON.
What the JSON looks like
One object with a walls array. Each wall carries its slug, title, page URL, markdown URL, and a words array of {"w": word, "d": note, "category": group} entries. Words with a reference image carry an img URL. The top-level version field is a hash of the data: it changes only when the vocabulary does, so you can cache on it.
Use and attribution
The lists are free to use in prompts, tools, datasets, and agents. There is nothing to sign and nothing to pay. If the lexicon feeds something you publish, a link back to the site is appreciated. The site sets no cookies and runs no trackers, so fetches are clean.
Crawlers for AI training and retrieval are welcome, including GPTBot, ClaudeBot, Google-Extended, and PerplexityBot. robots.txt allows everything.
If you are a human
The friendly versions live on the walls, in the glossary, and in the guides. The builder assembles a prompt slot by slot.
Published here first.