Build a small disposable experiment to answer a concrete engineering question without silently becoming production architecture.
Install
npx skillscat add sourabhjainr/auren/prototype Install via the SkillsCat registry.
SKILL.md
AUREN Prototype
Use this skill when uncertainty is best resolved by a runnable experiment rather than more discussion or production implementation.
Contract
A prototype has one explicit question, a bounded scope, a measurable success signal and a disposal or promotion decision.
Flow
question -> hypothesis -> minimal slice -> measure -> compare -> decide -> record
- Read the repository map and relevant contracts first.
- Define the hypothesis and the smallest experiment that can falsify it.
- Reuse canonical context, interfaces and test fixtures where practical.
- Keep prototype code isolated from production ownership unless promotion is explicit.
- Measure the agreed signal and capture inputs, output and limitations.
- Decide: discard, iterate once with a changed hypothesis, or promote through the normal implementation path.
Guardrails
- No new persistent repository graph, memory store or evidence store.
- No hidden production dependency from a prototype.
- No weakening of security, permission or verification gates.
- No benchmark claims without reproducible inputs and a receipt.
- Promotion requires normal implementation, tests, review and rollout gates.
Output
Record the question, hypothesis, experiment boundary, result, evidence identifiers, unknowns and promotion/disposal decision.