Migrate phrasing layer from Groq to Anthropic Claude (CLAUDE.md rule #9)

Fix still-forming cards mislabeling stable-but-noisy signals; add audio diagnostics

Phase B3 second slice: replace Mirror Feed with evidence-gated Home feed

Phase B3, You page: evidence-based standing portrait replaces trait scoring

Phase B2 first slice: per-signal evidence gating, replace population norms

Phase B1: quarantine pyannote/voiceprint off the live path, remove Upload UI

Log mic audio duration vs last transcribed timestamp; fix stale Whisper message

Switch ASR from Groq Whisper to Deepgram Nova-3

Route mic permission through a dedicated extension tab; parallelize mic+tab transcription

Capture the user's own mic as a second stream in Meet recording

Add job timeout watchdog, session pagination, feedback modal, and store assets

Add feedback endpoint with Resend email notifications

Fix backend round(dict) crash + tab capture permission + consent/results UX

Fix and update privacy policy

Remove enrollment gate, add 15-session monthly cap

LLM-generated portrait tags (max 4, behavior-specific)

Portrait layout, clean background, richer Recurring Themes

Fix missing trend signal in mirror feed — synthesizer was blind to numeric improvement

Richer portrait + collapsible mirror feed

Unlock behavioral vocabulary — break repetition loop across sessions

Remove reflect tab — saves one full LLM call per session

Major feature update: Google Meet recording, cumulative mirror feed, UX overhaul

Fix multi-speaker OOM + talk ratio inflation + add backend logging

Fix get_sessions 500 crash + extension panel stability

Add privacy policy page at /privacy

Add public /health endpoint for uptime monitoring

Rename to mirror., scroll reveal animations, waveform logo, deployment config

Profile intelligence, context-aware baselines, and Mirror Feed

Parallel transcription, SSE reliability fixes, HistoryView optimisation

Dimension scoring, expanded signals, and full results UI overhaul

Better diarization with pyannote, interruption detection, conversation summary with transcript

MVP working - audio analysis pipeline with Whisper, MFCC diarization, Groq insights