v1.0.4 — afterwords configure, requirements split, Red Dwarf gallery

What's new in v1.0.4

Added

• afterwords configure — new subcommand to toggle the Qwen3-1.7B model for the launchd service without editing plists by hand. afterwords configure --with-1.7b writes ~/.afterwords-server, regenerates the plist, and reloads launchd; --no-1.7b reverts. afterwords status shows 1.7B: enabled when set. bash setup.sh picks up the config on reinstall.
• Cursor IDE hook — ~/.claude/hooks/cursor-tts-hook.sh auto-configured by setup.sh when Cursor 1.7+ is detected. Fires on afterAgentResponse events; uses cursor as the agent key for .afterwords voice overrides.
• Red Dwarf gallery — 5 new voice families (Arnold Rimmer, Dave Lister, The Cat, Holly, Kryten) cloned from in-character BBC monologues. Brings the gallery to 103 families / 284 profiles.
• requirements-clone.txt — cloning-only dependencies (faster-whisper, noisereduce) split out of requirements.txt. setup.sh --server-only installs the server-only subset; clone-voice.sh checks for the clone deps and prints a helpful message if missing.

Changed

• holly reclone — new reference from the April Fool clip (pure solo speech, no crowd noise). Higher fidelity than the previous source.
• Architecture docs expanded — covers all six agent integrations and harness details for each.

Fixed

• tegan-jovanka synthesis garble — reference clip now splices only clean solo-speech windows, eliminating music bleed that caused Qwen3 to hallucinate phonemes.
• OG metadata voice count — updated from 275 → 284 after the Red Dwarf addition.