Cooking with constraints: A designer’s framework for better AI prompts

This article presents a designer's framework for writing more effective AI prompts, drawing an analogy between cooking preparation and prompt engineering.

• •Polite, vague prompts introduce ambiguity; models need clean instructions and clear constraints, not pleasantries
• •The TC-EBC framework (Task, Context, Elements, Behavior, Constraints) structures prompts to collapse uncertainty into actionable guidance
• •A vague natural-language prompt vs. a TC-EBC prompt in Figma Make demonstrates measurably richer, more intentional design output
• •Prompting is iterative: revision prompts should layer in context while explicitly constraining what the model should not change
• •Context engineering—curating what the model sees and weighs—keeps intent pure; providing Figma frames yields better results than screenshots because they carry structured data