FIX: OAuth PKCE code_verifier missing on callback

FIX: Prevent post-booking loop — one meeting per session

UX: Update Jessica greeting with Orpheus vocal direction tags

UX: Update welcome greeting to introduce Jessica by name

CHORE: Bump version to 1.7.10 — force fresh HF rebuild

FIX: Enable TTS without requiring microphone permission

CHORE: Bump version to 1.7.8

DOCS: Update CLAUDE.md with accurate architecture and commands

FIX: Replace decommissioned playai-tts with Groq Orpheus TTS model

FIX: Resume shared global AudioContext before TTS playback

FIX: Enable TTS without requiring microphone permission

FIX: Stop recording and handle farewell messages after booking

FIX: Extract availability responses from workflow agent output

FIX: Re-enable VAD and improve silence detection for faster response

CHORE: Bump version to 1.7.1

FIX: Preserve email from landing page when user info is saved

FEAT: Add privacy policy and terms of service for Google OAuth verification

FEAT: Restore availability checking for general calendar queries

BRANDING: Update description from "speechbot" to "Personal Voice Agent"

UX: Hide microphone UI during initial greeting

UI: Remove quick-action buttons for voice-only interface

FIX: Add loop detection to prevent repetitive asking for missing info

UX: Delay microphone start until TTS greeting finishes

FIX: Send session ID to /api/session-info for menu color updates

DEBUG: Add extensive logging for menu color update troubleshooting

HOTFIX: Disable noise filtering that was blocking all speech

FEAT: Add welcome greeting, faster response, and noise filtering

DEBUG: Add comprehensive logging for menu color updates

FIX: Resolve double booking_complete trigger timing issue

FIX: Remove auto-scroll + add debug logging for menu color updates

FIX: Move audio visualizer to bottom-right corner to avoid covering menu

UX: Fix recording popup covering menu items + reverse color coding

FIX: Update VOICECAL_LINKEDIN_DIAGRAM links from v2 to v3

FIX: Update LinkedIn diagram from voiceCal-ai-v2 to voiceCal-ai-v3

FIX: Preserve check_availability tool responses verbatim - show all time slots

FEAT: Add session reset after booking + celebration redirect

UX: Add LLM_DEBUG environment variable to suppress verbose logging

FIX: Update HuggingFace repo_id from v1 to v3 for secrets auto-update

FIX: Update OAuth redirect URI from v1 to v3

FIX: Default to Google Meet for all meetings

FIX: TTS outer circle visualization and mute button behavior

UI: Remove black background from audio visualizer

FEAT: Add dual-circle audio visualization for user and TTS

UI: Replace linear waveform with circular design (unmute.sh style)

UI: Replace linear waveform with circular design (unmute.sh style)

FIX: Strip 'assistant:' prefix breaking fireworks celebration

FIX: Topic extraction failing on common phrasings

FIX: Force LLM to use exact user_info values for booking parameters

DEBUG: Add comprehensive duration tracking logging

FIX: Handle space-separated times from STT (e.g., "1 20 PM")