← Back to Home

Authoring Methodology

What Build It Guides Are

A beginner-first, verification-driven instructional guide that transforms a working application into a step-structured learning system where each step:

Why This Approach is Distinctive

This is not "AI-generated code presented as-is." It is a structured production workflow that treats code as the executable reference and the guide as a validated instructional layer, with safeguards against drift, missing steps, and vague explanations.

Inputs

Roles in the Collaboration

0ne29 (author/producer): Owns system design, constraints, stage boundaries, QA expectations, and the "step contract" (what every step must include and prove). Also runs the early-stage stress testing and iteration loop to intentionally converge on a reliable format and workflow.
ChatGPT (GPT-5.2): Contributes early-stage structuring—template logic, decomposition strategy, boundary enforcement, and "explain + verify + recover" scaffolding.
Claude (Opus 4.5): Executes high-volume synthesis under constraints—consistent step-by-step prose, aligned code snippets, and internal consistency across the guide.

Method (Pipeline)

1
Stage freeze & constraints

Define what is in scope and what is explicitly out of scope.

2
Atomic decomposition

Break the system into "one step = one unit of change," including dependencies and architectural placement.

3
Instruction-as-verification

Embed checkpoints and "if something breaks" recovery notes so the guide doubles as a QA harness.

4
End-to-end validation

Consolidate into a runnable reference and validate that runtime behavior matches the guide's claims.

5
Feature verification suite

Add feature-by-feature verification checklists to confirm expected behavior and guard against regression.

Quality Controls

These controls prevent "LLM fluff" and ensure consistency:

Contracted step schema: Each step includes purpose, fit, code, explanation, why-it-matters, checkpoint, and recovery.
Dependency tracking: Each step declares prerequisites, preventing silent leaps.
Run-based validation: The final code is executed to confirm key behaviors (single window, timer accuracy, settings apply logic).
Scope enforcement: Explicit "does not include" constraints are asserted and re-verified near completion.

AI Collaboration Notes

Division of Labor

0ne29 (Human):

ChatGPT (GPT-5.2):

Claude (Opus 4.5):

Verification Standard

Guide assertions are backed by explicit verification checklists and execution of the reference code to confirm runtime behavior.

Note on Authorship

The core contribution is a repeatable pipeline that converts requirements into a verified, beginner-readable system with traceable rationale per change and per feature.

What This Demonstrates