Before you build anything, you see the full library structure and understand why each folder exists. CLAUDE.md is the constitution: it tells Claude who you are, how you work, and which files to load before you say a word.

• How to structure the 7-folder library: projects, wiki, frameworks, daily-logs, raw, drafts, CLAUDE.md
• What the CLAUDE.md constitution needs to say and how to co-evolve it over time
• Outcome: A working library with CLAUDE.md loading your context automatically at every session start

The frameworks folder is the part most people skip. You define the criteria. Claude applies them to every output it produces. This part builds one framework file for the most consequential quality call in your role.

• What makes a framework file different from a general instruction: pass/fail criteria, not guidance
• How to capture discipline-specific judgment so Claude applies your quality bar, not a generic one
• Outcome: One framework file encoding your highest-value quality criteria, applied in Part 04

Four commands give the system a daily heartbeat: one maps your priorities to your calendar each morning, one preps you before a meeting, one captures mid-day priority shifts, and one writes decisions to the library and builds tomorrow’s plan.

• How the morning planner matches priorities to calendar blocks by depth
• How the daily shutdown routine makes the system compound: each session close makes the next morning smarter
• Outcome: Four working commands and a daily rhythm that takes under 10 minutes per day

With the library seeded and the daily loop running, the draft command produces emails, LinkedIn posts, landing pages, and short video scripts grounded in your actual project context. Before saving, it runs your framework criteria and flags what to fix.

• Four output formats with built-in structural rules: email, LinkedIn post, landing page, short video script
• How the QA report diagnoses output at the criterion level before the draft is finalised
• Outcome: A working draft command and a first draft produced from your own project context

The library compounds when it stays honest. The governance check runs six specific checks once a week: contradictions, stale claims, orphan pages, missing concepts, neglected projects, and unsourced claims. Nothing changes until you confirm each fix.

• Six specific checks and what each one catches before it reaches a meeting or a brief
• The difference between a system that compounds and one that quietly drifts
• Outcome: A working governance check command and your first clean library run complete