Part I: Terms, Origins, and Paradigm Shifts

When a change truly begins to enter reality, what changes first is often not the tools, but the language. The old vocabulary is still there, yet people feel with increasing frequency that it is no longer sufficient. The problem clearly lies in knowledge entry points, verification loops, handoff structure, and responsibility boundaries, but teams still use prompt quality, the number of tools, and model strength to explain success and failure. Once language falls behind, solutions fall behind with it. That is the problem here: why do the old words begin to fail, why are new ones forced into being, and why does the center of gravity move from the performance of the model itself to the organization of the system? Before methodology appears, the coordinates must first be set. See Figures 1-1 through 1-4 in this part. Figure 1-1. The relationship among Prompt, Context, Agent, Workflow, and Harness graph LR P["Prompt engineering<br>Express intent clearly"] --> C["Context engineering&