← All terms Module 1: The Translation Layer

Three-Constraint Rule

cognitive-interface-architecture / three-constraint-rule

Definition

The principle that every well-formed prompt encodes three constraints: Intent (what you want the model to do), Context (the world it is operating in), and Guardrails (what it must never do). Missing any one produces predictable failure. The Three-Constraint Rule is the minimum viable architecture for every prompt.

What this prevents

A prompt with only Intent is vague. A prompt with only Context is accurate but useless. A prompt with only Guardrails is safe but empty. Most failed prompts answer only the first question: poorly.

Related terms

See all terms in Module 1: The Translation Layer →

See this term applied in production

The Agent Control Architecture Pack includes deployable system prompts, AGENTS.md templates, and fully-worked BYOP rebuilds that operationalise every precision term.

Get ACAP ($89) → Subscribe to The Constraint