Agentic Coding
in Practice
By Matthew Sinclair
A hands-on course on building real software with AI coding agents, in fifteen chapters over three levels: one human with one agent, then a practitioner's method at project scale, then one human running a fleet of agents. It is built from the full git histories of three projects, 22,945 commits up to 14 September 2026, and from one honest failure.
New here? Start Here — a short, free orientation. Or skip ahead to the syllabus or 101: The Landscape.
Fifteen chapters, three levels
The chapters are numbered the way undergraduate courses are. The first digit is the level, and each level builds on the one before. The first year, 1xx, is for someone new to agentic coding: one human and one agent, and the habits and mechanics that make a single session productive and safe. The second year, 2xx, is the practitioner with a method at project scale. The third year, 3xx, is multi-agent fleets.
1xx, first year
2xx, second year
- 201: Specification and the Work Unit
- 202: Rules, Skills and Critics
- 203: Structural Governance and Recovery
- 204: Instruments That Cannot Fail
- 205: Testing with Agents
- 206: Debugging and Diagnosis
- 207: Work Without a Test Suite
3xx, third year
- 301: The Step Up
- 302: The Whiteboard and the Shared Tree
- 303: Command, the Pen and the Human
- 304: Verification and Failure in the Fleet
Each chapter's hub links its theory, exercises and reference card, and the insights it homes. See it all at a glance on the course map.
What you'll get
Attendee Pack
The course companion: all fifteen chapters, each with its theory, exercises and reference card, ready to follow along.
Complete Reference
Everything in the Attendee Pack, plus the 71-insight reference, the Enterprise guide, and a PDF for every chapter.
Insights Reference
The 71 Agentic Coding Insights (ACIs): the field-tested rules, each with the evidence behind it, included in both bundles.