Optimizes LLM costs by delegating heavy tasks to a cheap worker model via a supervisor-worker pattern, compatible with multiple providers and IDE clients.
