The base-level practice of writing an explicit prompt so a coding agent doesn't have to infer missing context about the task itself. The foundation underneath Context Engineering in the AI-Native Developer Workflow.
Rather than a bare instruction like "implement task number one," the prompt spells out what the referent even is — e.g. "implement task number one, which is a GitHub issue" — so the agent doesn't need to guess the task's source or format.
Structuring a prompt with markdown — headers, bullet lists — so distinct sections (e.g. functional requirements, visual references, data instructions) are visibly separated for the model, rather than run together as prose. Used as the assembly format for a full-stack Context Engineering prompt that combines functional, visual, and data context into one document.