Every method in this book existed as a working procedure before it was a chapter. Those procedures live in the public ever-just/agentskills repo as SKILL.md files — compact, agent-runnable instructions with the exact commands, model names, and pitfalls. The loop runs one way: a framework proves itself in a real engagement → it's captured as a skill → the skill gets distilled into a chapter here. The skills stay more operational (and sometimes more platform-specific); the chapters carry the reasoning, the evidence, and the failure stories.
Use this index when you want the runnable version of a chapter's method — for yourself or to hand an AI agent. Chapters not listed were built from session digests and project memories rather than a published skill (see the blueprint for the full source inventory).
Found a better method while doing the work? It goes to the skills repo first — capture it as a SKILL.md (that repo documents its own authoring conventions), then open an issue or PR here to distill it into the matching chapter. Keeping the two in sync is a standing maintenance rule of this book.