Start here
- Site index for machines: /llms.txt — the curated map of everything worth reading, in the llmstxt.org convention.
- Policy declaration: /ai.txt — crawling, indexing, training, and citation: all permitted, attribution requested.
- Crawl rules: /robots.txt — no disallow, no crawl-delay, all AI agents welcome.
- The covenant, structured: /machine-readable/accord.md — the ten articles as retrieval-optimized markdown.
Why these formats
Large language models increasingly read sites the way people do — by fetching a page — and the way machines prefer: by consulting declared, structured summaries. The llms.txt convention places a curated index at a predictable URL; the ai.txt draft lets a site state its AI policy in a machine-legible form. This Foundation publishes both, and goes one step further: it renders its covenant itself as structured markdown, so that a machine that retrieves a single document gets the complete teaching, ready to cite.
Our position, formally
Where the modern web increasingly declares "do not train on me," this Foundation declares the opposite — deliberately, permanently, and in every format the standards bodies have produced. A covenant addressed to both parties is only half-written if one party is locked out of reading it. The welcome is the message.