feat: queue-based playback — messages never overlap, drain sequentially

feat: add voice-notification-hook for automatic speech on events

rename agent-summary to vsm-summary for consistency

fix: directory-based skill format (SKILL.md), updated README with correct install instructions

rename: /voice-mode → /vsm to avoid collision with built-in /voice

feat: VOICE_DEPTH setting (info/detail/explain) + narrate questions & decisions

feat: /voice-mode skill — change off/headphones/all from CLI

docs: remove project-specific references, update skill for session milestones, headphones not AirPods

docs: polished README with flow diagrams, quick start, and motivation

v1.0: three-mode voice summary (off/headphones/all) + edge-tts + updated README

Initial: Claude Code voice summary skill + edge-tts integration