Table of contents
Opening
Part 1 — Start from a Repository
Set up OpenCode, trace the Task API, refine a change, and diagnose a regression.
Part 2 — Shape and Document a Change
Break down an epic, map the system, discuss a proposal, and record an architecture decision.
Part 3 — Investigate a Production Incident
Use Grafana to check evidence from a simulated outage and write the postmortem.
Part 4 — Build the Harness
Add tests, merge checks, alerting, review, and agent controls for the problems the exercises surfaced.
- 11Building Your Harness: A Strategic Overview
- 12Focused Tests and Coverage Gates
- 13Required CI Checks and Merge
- 14Instrument the Service and Verify the Alert
- 15Prompt Engineering for Code Assistants
- 16Custom Agents & Sub-Agents
- 17AI-Assisted Code Review
- 18AGENTS.md & Project Context
- 19Repository Custom Instructions
- 20Agent Skills & Modular Expertise
- 21Hooks, Validation & Agent Automation
- 22MCP (Model Context Protocol)
- 23AI-First Security & Governance
Part 5 — Prove It on a Change
Measure the work and use the harness to ship searchable task labels in the lab.
Reference
This book is licensed under Creative Commons Attribution 4.0 International.