What problem does it solve? Teams often confuse acceptance tests with lower-layer checks like ROM conformance or graphics smoke tests, or write acceptance scenarios after implementation that merely rubber-stamp existing behavior. This Skill enforces a contract-first discipline: acceptance scenarios are authored before implementation, boot the real production composition, and stay red for the intended missing behavior until the feature lands. ## Core Features & Use Cases - Contract-first scenario authoring: Add or update acceptance scenarios before production code, verify they fail for the intended missing behavior, then freeze the contract so implementation cannot quietly weaken it. - Production composition enforcement: Boot the real non-rendering runtime through the shared acceptance harness instead of hand-assembling sessions, faking only true host boundaries like audio output, wall clock, and save roots. - Layer boundary discipline: Distinguish acceptance from ROM conformance and graphics smoke tests, stop before GPU rendering, and record precise exemptions for non-behavioral deliverables. - Use Case: When a deliverable changes dialogue flow or map transitions in the LÖVE runtime, author a scenario that drives semantic input through the production runtime, waits on semantic state, and asserts user-visible outcomes across each ready game version. ## Quick Start Ask the agent to author an acceptance scenario for the new dialogue flow deliverable using the production-composed acceptance harness before any implementation begins.
