Continuous improvement requires three things: a template to fill in, a procedure to follow, and an artifact to produce. Each cycle improves all three.
| Layer | What | Purpose |
|---|
| Template | What you fill in | Consistent structure |
| Procedure | How you fill it in | Repeatable process |
| Artifact | What you produce | Comparable result |
What templates exist and what question does each answer?
Locations
Content
| Template | Question |
|---|
Page Flow (.claude/rules/page-flow.md) | What rhythm does every page follow? |
Fact-Star (.claude/rules/fact-and-star-architecture.md) | Where does content live? |
Thinking
Business
Workflows
A workflow consumes one or more templates to produce a result artifact. More complex work uses more templates.
| Workflow | Question | Templates Used | Proven |
|---|
| Create PRD | Problem, for whom, measured how? | 5 maps + PRD (index, prompt-deck, spec) | 6+ PRDs |
| Industry Analysis | 5P scan of an industry? | 5P structure, Business Model Canvas | 8+ industries |
| Business Venture | Full persuasion loop? | Business Model Canvas, ICP, Work Chart | 3+ ventures |
| Compare Countries | Scored assessment on 25 dimensions? | ICP | 4+ countries |
| Meta Article | Idea to published piece? | Page Flow | 10+ articles |
| Fact-Star Migration | Hub/spoke restructuring? | Fact-Star | 5+ sections |
| DePIN Analysis | Token evaluation against thesis? | Business Model Canvas, Capability Map | 3+ tokens |
Skills
| Template | Skill | Question |
|---|
| Meta Article Template | copywriting | What shape does a /meta thought piece take? |
| Industry Analysis Template | deep-research | What 5P directory structure fits an industry? |
| ICP Output Template | create-icp | What defines a customer archetype? |
| Audit Loop Template | prompt-architecture | Did the skill execution improve the template? |
| Prediction Template | tracking-predictions | What did we predict, with what evidence? |
| Stage Templates | researching-business-ideas | What gate does each idea stage require? |
Context
Questions
What would change if you held every template to the same standard you hold code — defined inputs, expected outputs, and a test that can fail?
- Which templates in the inventory currently lack a feedback loop — and is that why they aren't compounding?
- When a template produces the same answer regardless of which system you point it at, is the template broken or is the operator?
- What's the difference between a template that improves your thinking and one that merely organises it?