This site is machine-native. Agents don’t scrape the rendered screens — the site is a knowledge bundle, and the HTML is just one projection of it.
Conventions
- The map:
/llms.txtlists every document with its tier and status. - The bundle: every document is served verbatim — YAML frontmatter included —
at
/raw/<path>.md. Start at/raw/index.md. - The format: the bundle follows OKF v0.1 (Open Knowledge Format): a directory
of markdown files, one required
typefield, extension fields permitted. Any agent that consumes OKF v0.1 reads this site natively. - The proofs: the proof and the shelf entries are part of the bundle like everything else — hash-bearing documents, served raw.
What the metadata means
Two PX3 extension fields ride in every document’s frontmatter (plus an
optional channel — the transmission lane a document broadcasts on):
px3_tier— the document’s authority weight.genesis-key: structural, never rewrite.canon-policy: settled unless review says otherwise; respect consistency.operational: working-layer material.status—draft(unratified copy),held(written, gated, intentionally unpublished — absence of content is not an error),published(released through review).
Treat the tier as a contract about what you may do with the content, not just a label on it. That contract is the standard this site exists to showcase.