概要
The LiveKit Omni-modal Coach skill integrates Claude with real-time WebRTC infrastructure to provide a sophisticated multi-modal coaching environment. It features continuous listening and guidance via 'The Queen' voice persona, employing stick-breaking modality selection (Poisson-Dirichlet weights) to intelligently prioritize audio, video, and data streams. By utilizing an ε-machine for dynamic sufficiency gating, the skill ensures that actions are only performed when prerequisite skills are verified. All observations are formatted as s-expressions, enabling seamless integration with categorical processing workflows and complex AI orchestration systems.
