A systematic tutorial refined from long-term AI coding practice, based on the "Implementation Planning" methodology. Related Skills: https://github.com/iannil/skills
Who This Book Is For
| Volume | Reader | Prerequisites | Goal |
|---|---|---|---|
| Volume 1: Getting Started | Developers new to AI coding | Basic programming concepts | Complete a small project independently using AI |
| Volume 2: Advanced | Experienced developers | Have used AI coding tools, encountered bottlenecks | Master systematic methods, improve efficiency and quality |
| Volume 3: Management | Technical managers | Responsible for team delivery | Establish team AI coding standards, manage risks |
How the Three Volumes Relate
- Volume 1 teaches you "how to do it"—follow the steps and build a working product.
- Volume 2 teaches you "why to do it this way"—understand the underlying principles, handle complex scenarios flexibly.
- Volume 3 teaches you "how to get the whole team doing it"—from individual capability to organizational capability.
Suggested reading order: If you are an individual developer, read Volume 1 first, then Volume 2. If you are a manager, read the overview section of Volume 3 first, then decide how to guide your team through Volumes 1 and 2.
About This Methodology
The methodology presented in this tutorial originates from the "Implementation Planning" approach (see zhurongshuo.com), and has been validated and iterated across a large body of real-world projects from the author's team—the engineering works and products listed below are its main carriers. These projects include enterprise management systems (unnamed drone system, strategic management system), self-developed products (AnyXMail, pinconsole, CodeCoder, Gojira), and dozens of experimental projects.
The Philosophical Foundation of This Methodology
This book extends the RC theoretical system (Process Realism: Observational Convergence and the Generation of Certainty) into the domain of software engineering. RC holds that the completeness of cognition lies solely in the capacity for process iteration, not in the finishedness of conclusions. Accordingly, this book frames AI engineering uniformly as progressive convergence among specification (blueprint and acceptance criteria), generation (AI coding), testing (inspection), and telemetry (progress and quality metrics)—the Six-Step Workflow, the three disciplines, and the fourteen skills are all implementations of this convergence mechanism at different project scales. One boundary must be stated explicitly: as prohibitive constraints, disciplines and standards can only rule out known bad paths; they cannot generate the single uniquely correct answer. Reading this book as a guarantee that "following the method necessarily leads to success" is inconsistent with RC's position, and with this book's.
Copyright
This book is original content protected by copyright. No part may be reproduced or distributed in any form without permission.
Terminology
This book uses the following terms:
- AI Coding Assistant: A code generation tool based on large language models, such as Claude Code, GitHub Copilot, etc.
- Blueprint: The architecture design document created before a project starts, i.e., the
CONTEXT.mdfile. - Inspection: A systematic check of AI-generated code to determine whether it meets the blueprint's expectations.
- Milestone: An independently verifiable functional unit, the basic building block of project progress.
See the appendix for the complete glossary.