ModelBreeder.com site discovery and publishing implementation notes

Purpose

This durable note records the v2.1.0 correction: external AI-ready guidance was previously converted into public guide pages about the guidance itself. That was the wrong target. The correct target is ModelBreeder.com as a plain PHP, source-backed educational site.

Correct implementation rule

The site should improve its own pages, templates, discovery files, metadata, route inventories, source evidence, and support boundaries. It should not copy external guidance into a public section, publish explainer pages about the reference, create doorway pages, hide machine-only instructions, insert model-manipulation text, invent source references, promise rankings, or claim certification.

Implemented target-site surfaces

• Canonical concept pages remain under /foundations, /theory, /architecture, /evolution-lab, /operations, /safety, /benefits, /blueprints, /tools, and /reference.
• Site evidence pages live under /site-evidence and describe the actual package: discovery files, canonical answer packets, metadata, source provenance, support boundaries, accessibility structure, search coverage, release validation, and review workflow.
• Public machine files remain advisory and target https://modelbreeder.com/.
• /docs remains the durable long-term source-memory layer.
• .uai/long-term-memory.uai remains a pointer-only ledger to /docs.
• The package remains root-extractable plain PHP with no database and no WordPress artifacts.

Owner and review context

• Owner: Michael Kappel.
• Public contact route: /contact.
• Package review date: 2026-06-27 UTC.
• Package target: plain vanilla PHP root site, not WordPress, not a CMS theme, not a plugin.

Release acceptance notes

Future agents should check for route drift, discovery-file drift, stale .uai memory, copied guidance pages, unsupported AI-ready claims, and package-target drift before producing a ZIP.