What problem does it solve? Single-agent coding and investigation workflows are slow and lack verification, so errors and weak solutions reach the user unchecked. This Skill races multiple parallel agents with competing approaches and adversarially verifies the winner before applying it. ## Core Features & Use Cases - Parallel Racing: Spawns 3-5 Flash workers (Trio or Squad mode) that implement genuinely different approaches to the same coding or investigation task simultaneously. - Blind Test Designer: Writes a full test suite without ever seeing any proposed solution, covering happy paths, edge cases, hostile inputs, and concurrency. - Tournament Selection & Breaker Agent: Scores competing solutions against weighted criteria, then a Breaker Agent runs hostile input fuzzing and mutation testing to destroy the winner before it ships. - Use Case: Ask it to fix a tricky bug or implement a feature; it races five coders, blind-tests the approaches, picks a winner, attacks it with a Breaker Agent, polishes any critical issues, and applies the verified solution to your project. ## Quick Start Run /turbo-boost on my task to refactor the authentication module and choose Squad mode when prompted.
