harness-programming
A short e-book coining **Harness Programming** - the discipline of composing the natural-language and configuration artifacts (prompts, skills, hooks, tools, memory) that shape how an AI agent behaves.
An in-progress short e-book coining and developing the term Harness Programming - the sub-discipline of harness engineering concerned with composing the natural-language and configuration artifacts (system prompts, AGENTS.md / CLAUDE.md, skills, hooks, tools, subagents, memory) that, together with an agent harness, produce a particular agent behavior. The framing draws a parallel: harness programming is to harness engineering what conventional programming is to software engineering. And just as programming diversified into schools (functional, logic, OO) each demanding its own runtime, natural-language programming is diversifying into schools that each require corresponding harness capabilities.
Active draft at roughly 29K words across a preface and ten chapters covering naming, harness-as-runtime, NL-as-code, schools, primitives, decomposition (skill vs hook vs tool), patterns, craft, outlook, and sources. A research/ corpus underneath grounds the chapters in an empirical catalog of program constructs extracted from the anthropics/skills set and the Playbooks DSL, which the book uses as the clearest concrete example of a structured-DSL school. Chapters are designed to be self-contained and readable out of order, with an ongoing structured review pass to sharpen each one.