One knowledge base. Many ways in. Six phases of one continuous chain — from why we build to whether we served. Crafted for the new hire reading on day one, the senior practitioner returning for depth, and the AI agent fetching at scale.
This is not a methodology. It is a working corpus. The chain it describes — why we build · before we build · what we shape · as we build · after we build · did we serve — runs in the work, every cycle. The pages here are the artefacts that make the chain learnable.
Six phases of one continuous chain. The first four run per cycle, in sequence. The last two are persistent — Why We Build sets the bet above any single cycle; Did We Serve is the relationship that holds across cycles.
Same knowledge, different depths. Pick the layer that matches the moment you arrived.
| Depth | When | Form |
|---|---|---|
| Glance | "I'm about to act" | Checklist |
| Skim | "What's this in 2 minutes?" | Field card or template |
| Do | "Walk me through it now" | Practice page |
| Diagnose | "What does going wrong look like?" | Clinic |
| Internalise | "Read it like a book" | Cycle phase |
| Grow | "How do I become this?" | Skill path |
| Believe | "Why this stance, not another?" | Principle |
The corpus exposes structured artefacts at the site root:
/llms.txt — index with one-line descriptions and a when to use which slice guide/llms-canon.txt — the cycle (canon), one file/llms-practice.txt — practice and clinics/llms-reference.txt — areas, glossary, templates, checklists/llms-roles.txt · /llms-skills.txt — role hubs and learning paths/llms-full.txt — everything, one file/corpus-graph.json — every page as a node, with derives_from, owners, roles, areas — for structured retrieval