What problem does it solve? Large epics split across multiple parallel AI sessions need a single coordinator that plans the work into waves, launches batches, answers questions against the actual code, accepts returns, and keeps a durable ledger — without this role, parallel sessions drift, duplicate work, or deadlock on merges. ## Core Features & Use Cases - Wave planning and dispatch: Splits an epic into batches, launches them through hcb-dev:wave-dispatch, and opens each next wave only when its gate clears. - Event-driven coordination loop: Handles start reports, questions, statuses, returns, landings, and tracker edits per a defined protocol, recording every event in a wave ledger before moving on. - Acceptance and findings ruling: Accepts batch returns per the order-return protocol, rules tracker candidates through hcb-dev:findings-pass, and manages the merge queue without ever merging itself. - Use Case: You assign a session "ты — мастер-сессия" over a milestone; it files an umbrella issue, draws the wave plan, launches parallel worker sessions, coordinates their questions and landings, and closes the epic with a final report. ## Quick Start Tell the session to act as the master session for your epic or milestone and let it plan the waves and coordinate the parallel batches.
