Prototype

Prototype

Create the smallest runnable experiment that can produce evidence for one decision. Keep the answer; do not let experimental code silently become production code.

Establish the Question

Write one falsifiable question and the evidence that would answer it.

Classify it:

• Logic or state: determine whether rules, transitions, invariants, or interactions behave coherently.
• Interface or UI: compare materially different interaction or presentation choices.
• Integration feasibility: verify that a library, protocol, platform, or boundary can support the required behavior.

Do not start until the prototype has one primary question. If several decisions are independent, split them into separate experiments.

Respect the Prototype Contract

• Do not create a prototype during a plan-only request unless the user explicitly authorizes the experiment.
• Use the operating system's temporary directory by default.
• Place it in the workspace only when access to the real build, route, components, or configuration is necessary and the task authorizes changes.
• Name and label workspace artifacts clearly as prototypes.
• Do not create branches, commits, dependencies, deployments, external resources, or persistent data without explicit authorization.
• Use scratch data and reversible local state.
• Do not modify production behavior to support the experiment.

Reuse the Existing Environment

Inspect the repository's runtime, package manager, components, fixtures, conventions, and available commands. Reuse existing dependencies and tooling.

Avoid installing a framework or building infrastructure for a question that a small script, local route, test harness, or temporary component can answer.

Provide one command that starts or runs the experiment.

Build Only the Required Fidelity

Make the prototype complete and functional for the question it answers. Exclude production concerns that do not affect the result, such as:

• generalized abstractions;
• exhaustive error handling;
• persistence unrelated to the decision;
• broad test coverage;
• production observability;
• compatibility beyond the evaluated path;
• visual polish unrelated to the UI choice.

Do not use placeholders for behavior that the decision depends on.

Prototype Logic and State

• Exercise difficult transitions, invalid inputs, repeated actions, concurrency, and recovery when relevant.
• Show the full relevant state after each action.
• Make assumptions and invariants visible in the experiment output.
• Compare alternative state models using the same scenarios.
• Prefer a small interactive CLI or deterministic harness over a large application shell.

Prototype Interfaces and UI

• Produce at least two materially different options when the question is comparative.
• Keep options reachable from one simple entry point.
• Reuse the project's design system when visual fidelity affects the decision.
• Make important states visible, including loading, empty, error, success, and constrained layouts when relevant.
• Optimize for learning, not decoration.

Prototype Integration Feasibility

• Verify the exact installed or documented versions.
• Exercise the real boundary with the smallest safe input.
• Capture response shape, errors, limits, latency, and lifecycle behavior that affect the decision.
• Avoid live writes, paid actions, or production systems unless the user explicitly authorizes them.

Evaluate the Evidence

Run the experiment and record:

• the question;
• the tested scenarios or alternatives;
• the observed result;
• the decision supported by the evidence;
• limitations that prevent generalizing the result;
• what must change in the production design.

Do not call the experiment successful merely because it runs.

Close the Experiment

• Keep temporary artifacts long enough for the user to inspect them.
• Report their absolute location and the run command.
• Do not merge or copy prototype code into production automatically.
• If implementation follows, transfer the validated decision and rebuild production-quality code under the normal implementation and verification workflows.
• Remove workspace prototype artifacts only when the user requests cleanup or the authorized workflow explicitly makes them disposable.

Completion Standard

Finish only when the prototype runs, directly exercises the material question, produces interpretable evidence, and remains clearly separated from production code and state.