What problem does it solve? It prevents unfinished or unverified implementation work by enforcing that each task from a delivery plan is coded strictly to its description, tested by actually running the tests, and documented in a shared test report instead of being taken on faith. ## Core Features & Use Cases - Single-task implementation: Takes exactly one task from the plan, checks the test report history to find the first unclosed task, and implements it without scope creep or unsolicited refactoring. - Evidence-based testing: Writes end-to-end, unit, and regression tests, runs them, and distinguishes real proof from mock/stub-only coverage of production-relevant branches. - Cumulative test report: Maintains one test_report file per slice with an up-to-date acceptance criteria coverage table, changed files list, and per-task history. - Use Case: A game development team has a task plan for a playable slice; the skill picks the next open task, implements the board-rendering logic against the agreed wireframes, runs the full test suite, updates the coverage table, and commits before handing off to code review. ## Quick Start Ask the agent to implement the next open task from the plan and update the test report with the results.
