constraint-specification

Constraint Specification

Constraints are the rules that shape AI output — what format to use, how long to be, what to include, what to exclude. Well-specified constraints produce predictable, useful outputs. Vague constraints produce inconsistent results.

Types of Constraints

Format constraints:

• Output structure (JSON, markdown, plain text, bullet points, prose)
• Section headings and organisation
• Required fields and optional fields
• Data types and schemas Length constraints:
• Word count ranges (not exact numbers — models are bad at counting)
• Section length proportions ("spend 60% on analysis, 40% on recommendations")
• Minimum and maximum bounds
• Conciseness directives ("be brief" vs. "be thorough") Content constraints:
• Topics to include and exclude
• Required information elements
• Prohibited content
• Source restrictions (only use provided context, don't use external knowledge) Tone constraints:
• Formality level
• Emotional register
• Audience-appropriate language
• Voice and style guidelines Quality constraints:
• Accuracy requirements ("cite sources", "flag uncertainty")
• Completeness requirements ("address all aspects of the question")
• Originality requirements ("don't repeat the question back")
• Actionability requirements ("every recommendation must be implementable")

Writing Effective Constraints

• Be specific: "Keep responses under 200 words" beats "be concise"
• Prioritise: When constraints conflict, state which wins. "Accuracy over brevity."
• Provide examples: Show what a constrained output looks like
• Test boundaries: What happens at the edge of each constraint?
• Separate hard and soft constraints: Hard constraints must always be met. Soft constraints are preferences.

Constraint Interactions

Constraints interact and can conflict:

• "Be thorough" vs. "Keep it under 100 words"
• "Be creative" vs. "Follow this exact format"
• "Be helpful" vs. "Don't give medical advice" Resolve conflicts explicitly in the prompt. Don't make the model guess which constraint takes priority.

Design Artefacts

• Constraint specification documents per output type
• Constraint priority hierarchies
• Constraint test cases (inputs designed to stress each constraint)
• Constraint violation examples (what bad looks like)
• Constraint evolution logs (how constraints changed and why)