Interactive Prototype

Create the smallest runnable artifact that exposes the target behavior and preserves the product decisions already made.

Before building

Require or infer a precise prototype question, audience, required fidelity, critical states, evidence plan, and stop rule. If these are missing and would materially change the artifact, route to prototype-question first.

Inspect available product context, conceptual model, design system, existing code, and prototype tools. Reuse an existing artifact when it can answer the question with a smaller change. Choose an available implementation method; do not require a particular external engine.

Build for evidence

• Implement only the critical path and states required by the question.
• Make loading, empty, error, success, permission, and recovery states real when they affect the decision.
• Label simulated data or behavior; isolate it from production systems.
• Preserve the user's confirmed terminology, constraints, hierarchy, and interaction intent.
• Add only the minimum instrumentation or observation hooks needed for the evidence plan.
• Keep the artifact easy to reset and share with intended reviewers or participants.

Handoff

Return the runnable artifact or implementation path, how to operate it, what is simulated, known limitations, the evidence to collect, and the stop rule. After testing, route observations to learning-loop-readout rather than treating completion as validation.

Boundaries

• Do not silently expand a prototype into production architecture.
• Do not add accounts, payments, destructive actions, live customer data, or external publication without explicit authorization.
• Do not polish unrelated surfaces after the prototype can already answer its question.
• Do not claim usability, desirability, or feasibility without observed evidence.