How to read: the course’s backbone, from left to right. The specification says what’s correct; the harness defines what the agent can do; the loop (in cyan) runs and corrects itself until it reaches the stopping condition; the quality gates decide whether the change can be released. Each track below explores one of these boxes in more depth.
The 6 tracks
🧭 AI-First Fundamentals
From autocomplete to agent—and why the fundamentals are deciding everything again. It ends with Spec-Driven Development.
⚙️ Harness & Loop
The system around the model: tools, context, permissions, sandbox—and execution cycles that know when to stop.
🔁 Workflow & Quality
Events that trigger agents, independent AI review, and the gates that decide whether the change moves forward.
🏛️ Architecture in the age of AI
Legacy without surprises, repositories agents can understand, and the fight against overengineering and AI-generated technical debt.
🛡️ Playbooks, Skills, and Governance
Reusable processes, skills, and rules that limit arbitrary decisions, and the 5 levels of autonomy.
🚀 Hands-On Exercise
You run the entire workflow in a real repository: spec → harness → loop → implementation → gates → review → delivery.
How to study this course
✓ What to DO
- ✓Follow T1 → T6 in order: each track uses the vocabulary from the previous one.
- ✓Mark each section as read—that’s how your progress advances.
- ✓Actually run the copyable prompts and commands in one of your repositories.
- ✓Mark anything unresolved as “I have a question” and review it in “My journey.”
✗ What NOT to do
- ✗Jump straight to the practice track without understanding the spec and harness.
- ✗Test the exercises directly on the main branch of a production system.
- ✗Treating the course as a list of tools — what lasts is the method, not the product of the moment.
- ✗Clear your browser data without exporting your progress in "My journey".
What you take away at the end
After completing the six tracks, you’ll have a ready-to-use toolkit for your team—not just notes: