Chapter 14: Ten Axioms of Agentic Development

Overview

This chapter bridges Parts 1-3 (experiential learning) with Parts 4-6 (technical implementation), formalizing the developer methodology that has been implicit throughout your work so far. Through General Agents, Applied Workflows, and SDD-RI, you have already practiced these axioms without naming them. Now we make them explicit, providing the rigorous foundation that governs all subsequent Python programming and agent-building work.

The Ten Axioms are not abstract theory. They are distilled from real agentic development practice and directly extend the Seven Principles introduced in Chapter 4.

Relationship to the Seven Principles (Chapter 4)

Dimension	Seven Principles	Ten Axioms
Audience	ALL users of AI agents	DEVELOPERS building agents
Scope	Mindset and approach	Methodology and practice
Level	How to think about agent work	How to engineer agent systems
Introduced	Part 1 (foundations)	Part 4 (technical bridge)

The Principles tell you how to think. The Axioms tell you how to build.

The Ten Axioms

#	Axiom	Principle Connection	Core Teaching
I	Shell as Orchestrator	P1: Bash is the Key	Shell coordinates all tools; programs compute
II	Knowledge is Markdown	P5: Persisting State	Markdown is the universal knowledge format
III	Programs Over Scripts	P2: Code as Interface	Production requires typed programs with tests and CI
IV	Composition Over Monoliths	P4: Decomposition	Build from composable, focused units
V	Types Are Guardrails	P6: Constraints	Type systems prevent errors before runtime
VI	Data is Relational	P5: Persisting State	SQL as default for structured data
VII	Tests Are the Specification	P3: Verification	TDG: tests define correctness
VIII	Version Control is Memory	P5: Persisting State	Git as persistent memory layer
IX	Verification is a Pipeline	P3: Verification	CI/CD automates verification
X	Observability Extends Verification	P7: Observability	Runtime monitoring extends testing

Layer Progression

Lessons	Layer	Focus
01-04	L1 (Manual)	Introduce axioms conceptually with concrete examples
05-07	L2 (Collaboration)	Explore axioms with AI assistance
08-10	L2 to L3	Axioms compose into integrated systems

Prerequisites

• Part 1: General Agents Foundations (Chapters 1-4)
• Part 2: Applied General Agent Workflows (Chapters 5-10)
• Part 3: SDD-RI Fundamentals (Chapters 11-13)

What You'll Learn

• How the shell serves as an orchestration layer distinct from computation
• Why markdown is the universal knowledge format for agent workflows
• The difference between disposable scripts and production programs
• How composition, types, and relational data create robust agent systems
• Test-Driven Generation (TDG) as the specification mechanism for AI-generated code
• Git as persistent memory and CI/CD as automated verification
• How observability extends verification from build-time into runtime