claude-voice 1.5.0 → 1.5.2

package/README.md

@@ -2,453 +2,122 @@
 
 [![npm version](https://img.shields.io/npm/v/claude-voice)](https://www.npmjs.com/package/claude-voice)
 [![License: PolyForm Noncommercial](https://img.shields.io/badge/License-PolyForm%20Noncommercial-red.svg)](LICENSE)
-[![Node.js](https://img.shields.io/node/v/claude-voice)](https://nodejs.org)
 [![Platform](https://img.shields.io/badge/platform-macOS%20%7C%20Linux-blue)]()
 
-Voice interface for Claude Code CLI. Speak commands, hear responses.
-
-<!-- Demo GIF placeholder - replace with actual recording -->
-<!-- ![Demo](docs/demo.gif) -->
-
-**Features:**
-
-- Speaks Claude's responses aloud (Text-to-Speech)
-- Transcribes your voice commands (Speech-to-Text)
-- Hands-free with wake word detection ("Hey Jarvis")
-- Works offline with local providers - no API keys required
-- Deep integration with Claude Code via hooks system
-
-## Quick Start
+Voice interface for Claude Code. Speak commands, hear responses.
 
 ```bash
 npm install -g claude-voice
 ```
 
-That's it! The extension auto-starts when you open Claude Code. Say **"Hey Jarvis"** followed by your command.
-
-**Upgrade voice quality:**
-- Better TTS: `claude-voice local --download` (Piper neural voices)
-- Best quality: `claude-voice openai` (requires API key)
-- Customize: `claude-voice setup`
+Say **"Hey Jarvis"** followed by your command. The extension auto-starts with Claude Code.
 
 ## How It Works
 
 ```
-You speak              Claude Voice            Claude Code
-    |                       |                       |
-    |--- "Jarvis..." -----> |                       |
-    |    (wake word)        |--- transcribe ------> |
-    |                       |      (STT)            |
-    |                       |                       |
-    |                       | <---- response ------ |
-    | <-- speaks aloud ---- |                       |
-    |        (TTS)          |                       |
+You speak → "Hey Jarvis..." → Wake word detected → STT transcribes → Claude Code receives
+Claude responds → Hook captures → TTS speaks aloud → You hear the response
 ```
 
-**Claude Code Integration:**
-
-| Hook | Purpose |
-|------|---------|
-| `session-start` | Auto-starts daemon when Claude Code launches |
-| `stop` | Speaks responses when Claude finishes |
-| `post-tool-use` | Announces tool completions (file reads, bash commands) |
-| `notification` | Voice alerts for permission prompts |
+The extension integrates via Claude Code hooks: auto-start on session, speak responses, announce tool completions, and voice alerts for permission prompts.
 
 ## Providers
 
-Choose local (free, offline) or cloud providers:
-
-| Capability | Local (Free) | Cloud |
-|------------|--------------|-------|
-| Text-to-Speech | Piper, macOS Say, espeak | OpenAI TTS, ElevenLabs |
-| Speech-to-Text | Sherpa-ONNX | OpenAI Whisper |
-| Wake Word | openWakeWord, Sherpa-ONNX | Picovoice Porcupine |
-
-<details>
-<summary><strong>TTS Providers</strong></summary>
-
-### Piper (Default)
-
-Local neural TTS with high-quality voices. No API key required.
-
-```bash
-claude-voice voice list                    # See available voices
-claude-voice voice download en_US-amy-medium
-claude-voice config set tts.provider=piper
-```
-
-### macOS Say
-
-Built-in macOS speech synthesis.
-
-```bash
-claude-voice voices                        # List available voices
-claude-voice config set tts.provider=macos-say
-claude-voice config set tts.macos.voice=Samantha
-```
-
-### OpenAI TTS
-
-High-quality neural voices. Requires `OPENAI_API_KEY`.
-
-```bash
-echo "OPENAI_API_KEY=sk-..." >> ~/.claude-voice/.env
-claude-voice config set tts.provider=openai
-claude-voice config set tts.openai.voice=nova
-```
-
-### ElevenLabs
-
-Premium voice synthesis. Requires `ELEVENLABS_API_KEY`.
-
-```bash
-echo "ELEVENLABS_API_KEY=..." >> ~/.claude-voice/.env
-claude-voice config set tts.provider=elevenlabs
-```
-
-</details>
-
-<details>
-<summary><strong>STT Providers</strong></summary>
-
-### Sherpa-ONNX (Default)
-
-Local Whisper models. No API key required. Supports 100+ languages.
-
-```bash
-claude-voice model list                    # Available models
-claude-voice model download whisper-small  # Best accuracy (488MB)
-claude-voice config set stt.provider=sherpa-onnx
-claude-voice config set stt.language=en    # or: tr, de, fr, es, etc.
-```
-
-| Model | Size | Speed | Accuracy |
-|-------|------|-------|----------|
-| whisper-tiny | 75 MB | Fast | Good |
-| whisper-base | 142 MB | Medium | Better |
-| whisper-small | 488 MB | Slower | Best |
-
-### OpenAI Whisper
+| | Local (Free) | Cloud |
+|---|---|---|
+| **TTS** | macOS Say, Piper, espeak | OpenAI, ElevenLabs |
+| **STT** | Sherpa-ONNX Whisper | OpenAI Whisper |
+| **Wake Word** | openWakeWord, Sherpa-ONNX | Picovoice |
 
-Cloud transcription. Requires `OPENAI_API_KEY`.
+**Quick presets:**
 
 ```bash
-claude-voice config set stt.provider=openai
-```
-
-</details>
-
-<details>
-<summary><strong>Wake Word Providers</strong></summary>
-
-### openWakeWord (Default)
-
-Purpose-trained wake word detection with high accuracy. Requires Python 3. Installed automatically during setup.
-
-```bash
-claude-voice openwakeword --install            # Install/configure openWakeWord
-claude-voice config set wakeWord.provider=openwakeword
-```
-
-Available models: `hey_jarvis` (default), `alexa`, `hey_mycroft`, `hey_rhasspy`
-
-### Sherpa-ONNX KWS (Fallback)
-
-Local keyword spotting, no Python required. Used as fallback when Python is not available.
-
-```bash
-claude-voice config set wakeWord.provider=sherpa-onnx
-claude-voice config set wakeWord.keyword=jarvis   # or: claude
-```
-
-### Picovoice Porcupine
-
-High-accuracy wake word detection. Requires `PICOVOICE_ACCESS_KEY`.
-
-1. Get a free access key at [Picovoice Console](https://console.picovoice.ai/)
-2. Configure:
-
-```bash
-echo "PICOVOICE_ACCESS_KEY=..." >> ~/.claude-voice/.env
-claude-voice config set wakeWord.provider=picovoice
-claude-voice config set wakeWord.keyword=jarvis   # jarvis, computer, alexa, etc.
+claude-voice setup              # Interactive setup wizard
+claude-voice openai             # Cloud TTS + STT (requires API key)
+claude-voice local --download   # Piper TTS + larger Whisper model (offline)
 ```
 
-Built-in keywords: jarvis, computer, alexa, americano, blueberry, bumblebee, grapefruit, grasshopper, hey google, hey siri, ok google, picovoice, porcupine, terminator
-
-</details>
-
 ## Configuration
 
-Config file: `~/.claude-voice/config.json`
-
 ```bash
-claude-voice config                        # View all settings
-claude-voice config get tts.provider       # Get specific value
+claude-voice config                         # View all
 claude-voice config set tts.provider=openai # Set value
-claude-voice config edit                   # Open in editor
-claude-voice config reset                  # Reset to defaults
+claude-voice config set stt.language=tr     # Change language
+claude-voice config edit                    # Open in editor
 ```
 
-<details>
-<summary><strong>TTS Options</strong></summary>
-
-| Option | Default | Description |
-|--------|---------|-------------|
-| `tts.provider` | `piper` | piper, macos-say, openai, elevenlabs, espeak, disabled |
-| `tts.autoSpeak` | `true` | Automatically speak Claude's responses |
-| `tts.maxSpeechLength` | `5000` | Maximum characters to speak |
-| `tts.skipCodeBlocks` | `true` | Skip code blocks when speaking |
-
-</details>
+Config file: `~/.claude-voice/config.json`
 
 <details>
-<summary><strong>STT Options</strong></summary>
+<summary><strong>All options</strong></summary>
 
 | Option | Default | Description |
 |--------|---------|-------------|
+| `tts.provider` | `macos-say` | macos-say, piper, openai, elevenlabs, espeak, disabled |
+| `tts.autoSpeak` | `true` | Auto-speak Claude responses |
+| `tts.maxSpeechLength` | `5000` | Max characters to speak |
 | `stt.provider` | `sherpa-onnx` | sherpa-onnx, openai, whisper-local, disabled |
-| `stt.language` | `en` | Language code (en, tr, de, fr, es, ja, zh, etc.) |
-
-</details>
-
-<details>
-<summary><strong>Wake Word Options</strong></summary>
-
-| Option | Default | Description |
-|--------|---------|-------------|
+| `stt.language` | `en` | Language code (en, tr, de, fr, es, ja, zh...) |
 | `wakeWord.enabled` | `true` | Enable wake word detection |
-| `wakeWord.provider` | `openwakeword` | openwakeword, sherpa-onnx, or picovoice |
-| `wakeWord.keyword` | `jarvis` | Wake word: jarvis, claude, computer, etc. |
+| `wakeWord.provider` | `openwakeword` | openwakeword, sherpa-onnx, picovoice |
 | `wakeWord.sensitivity` | `0.5` | Detection sensitivity (0.0-1.0) |
-| `wakeWord.playSound` | `true` | Play sound on detection |
-
-</details>
-
-<details>
-<summary><strong>Voice Output Options</strong></summary>
-
-When enabled, Claude formats responses with a spoken abstract before technical details.
-
-| Option | Default | Description |
-|--------|---------|-------------|
-| `voiceOutput.enabled` | `false` | Enable TTS-friendly formatting |
-| `voiceOutput.abstractMarker` | `<!-- TTS -->` | Marker separating spoken/technical content |
-| `voiceOutput.maxAbstractLength` | `200` | Max characters for spoken abstract |
-
-```bash
-claude-voice output enable                 # Enable voice-friendly formatting
-claude-voice output status                 # Check current status
-```
-
-</details>
-
-<details>
-<summary><strong>Tool Announcements</strong></summary>
-
-| Option | Default | Description |
-|--------|---------|-------------|
+| `voiceOutput.enabled` | `false` | TTS-friendly response formatting |
 | `toolTTS.enabled` | `false` | Announce tool completions |
-| `toolTTS.mode` | `summarize` | summarize or completion |
-| `toolTTS.announceErrors` | `true` | Announce tool errors |
-
-</details>
-
-<details>
-<summary><strong>Keyboard Shortcut</strong></summary>
-
-| Option | Default | Description |
-|--------|---------|-------------|
-| `shortcut.enabled` | `false` | Enable keyboard shortcut |
-| `shortcut.key` | `CommandOrControl+Shift+Space` | Key combination |
-
-**Modifiers:** CommandOrControl, Command, Control, Shift, Alt
-
-</details>
-
-<details>
-<summary><strong>Recording Options</strong></summary>
-
-| Option | Default | Description |
-|--------|---------|-------------|
-| `recording.sampleRate` | `16000` | Audio sample rate (Hz) |
-| `recording.silenceThreshold` | `2500` | Silence duration to stop (ms) |
-| `recording.silenceAmplitude` | `500` | Amplitude threshold |
+| `recording.silenceThreshold` | `3500` | Silence duration to stop recording (ms) |
 | `recording.maxDuration` | `60000` | Max recording length (ms) |
 
 </details>
 
-<details>
-<summary><strong>Server Options</strong></summary>
-
-| Option | Default | Description |
-|--------|---------|-------------|
-| `server.port` | `3456` | HTTP server port |
-| `server.host` | `127.0.0.1` | Server host |
-
-</details>
-
 ## CLI Commands
 
 ```bash
-# Daemon Management
-claude-voice start              # Start daemon
-claude-voice stop               # Stop daemon
-claude-voice restart            # Restart daemon
-claude-voice status             # Check status
-
-# Setup
-claude-voice setup              # Interactive setup wizard
-claude-voice doctor             # Diagnose issues
+# Daemon
+claude-voice start / stop / restart / status
 
-# Provider Presets (Quick Configuration)
-claude-voice openai             # Use OpenAI for TTS and STT (cloud)
-claude-voice openai --voice nova # Specify voice (nova, alloy, echo, fable, onyx, shimmer)
-claude-voice local              # Use Piper TTS + Sherpa-ONNX STT (offline)
-claude-voice local --download   # Configure and download required models
-claude-voice openwakeword       # Switch to openWakeWord (better wake word detection)
-claude-voice download-models    # Download models for current config
+# Setup & Diagnostics
+claude-voice setup                # Interactive wizard
+claude-voice doctor               # Diagnose issues
 
 # Models & Voices
-claude-voice model list         # List STT models
-claude-voice model download <id>
-claude-voice voice list         # List TTS voices
-claude-voice voice download <id>
+claude-voice model list / download <id>     # STT models (whisper-tiny/base/small)
+claude-voice voice list / download <id>     # Piper TTS voices
 
-# Hooks
-claude-voice hooks install      # Install Claude Code hooks
-claude-voice hooks status       # Check installation
+# Wake Word
+claude-voice openwakeword --install         # Better wake word detection
 
 # Testing
-claude-voice test-tts "Hello"   # Test text-to-speech
-claude-voice test-stt file.wav  # Test speech-to-text
+claude-voice test-tts "Hello"
+claude-voice test-stt recording.wav
 
 # Utilities
-claude-voice logs               # View daemon logs
-claude-voice logs -f            # Follow logs
-claude-voice devices            # List audio devices
+claude-voice logs -f              # Follow daemon logs
+claude-voice devices              # List audio devices
 ```
 
-Run `claude-voice --help` for all 50+ commands.
-
 ## Platform Support
 
-| Feature | macOS | Linux |
-|---------|-------|-------|
-| TTS | Piper, Say, OpenAI, ElevenLabs | Piper, espeak, OpenAI, ElevenLabs |
+| | macOS | Linux |
+|---|---|---|
+| TTS | Say, Piper, OpenAI, ElevenLabs | espeak, Piper, OpenAI, ElevenLabs |
 | STT | Sherpa-ONNX, OpenAI | Sherpa-ONNX, OpenAI |
 | Wake Word | openWakeWord, Sherpa-ONNX, Picovoice | openWakeWord, Sherpa-ONNX, Picovoice |
-| Keyboard Shortcut | Cmd+Shift+Space | Ctrl+Shift+Space |
-| Terminal Injection | AppleScript | xdotool (X11), dotool (Wayland) |
 
-**Requirements:**
-- Node.js 18+
-- Microphone access
-- Python 3 (for openWakeWord; falls back to Sherpa-ONNX if unavailable)
+**Requires:** Node.js 18+, microphone access. Python 3 recommended (for openWakeWord).
 
 ## Troubleshooting
 
-Run diagnostics:
-
-```bash
-claude-voice doctor
-```
-
-<details>
-<summary><strong>Common Issues</strong></summary>
-
-**Daemon won't start**
 ```bash
-claude-voice logs              # Check logs
-claude-voice start -f          # Run in foreground for debugging
-```
-
-**No audio output**
-```bash
-claude-voice test-tts "Hello"
-claude-voice config get tts.provider
+claude-voice doctor               # Auto-diagnose and fix issues
+claude-voice logs                 # Check daemon logs
+claude-voice start -f             # Run in foreground for debugging
 ```
 
-**Wake word not detecting**
-- Upgrade to openWakeWord: `claude-voice openwakeword --install`
-- Check microphone permissions in System Preferences
-- Run `claude-voice devices` to verify microphone
-- Adjust sensitivity: `claude-voice config set wakeWord.sensitivity=0.7`
-
-**Text not appearing in terminal**
-- macOS: Allow Terminal in System Preferences > Privacy > Accessibility
-- Run `claude-voice doctor` to check terminal injection status
-
-</details>
-
-## API Reference
-
-<details>
-<summary><strong>HTTP API (port 3456)</strong></summary>
-
-| Endpoint | Method | Description |
-|----------|--------|-------------|
-| `/status` | GET | Daemon status and provider info |
-| `/tts` | POST | Speak text `{"text": "...", "priority": false}` |
-| `/tts/stop` | POST | Stop current playback |
-| `/stt` | POST | Transcribe audio (multipart/form-data) |
-| `/config` | GET | Get configuration |
-| `/config` | POST | Update configuration |
-
-</details>
-
-## Changelog
-
-### v1.5.0
-- **openWakeWord as default** - Purpose-trained "Hey Jarvis" model (200K+ samples) replaces generic Sherpa-ONNX KWS for much better wake word accuracy
-- **Smart provider detection** - Automatically uses openWakeWord when Python 3 is available, falls back to Sherpa-ONNX KWS otherwise
-- **espeak TTS provider** - Linux native TTS support
-- **Setup wizard upgrade** - All presets now prefer openWakeWord, Custom preset includes wake word provider choice
-
-### v1.4.0
-- **License change** - Switched to PolyForm Noncommercial (commercial use restricted)
-
-### v1.3.19
-- **`shh` command** - Stop TTS instantly
-
-### v1.3.18
-- **`listen` command** - Manual voice trigger without wake word
-
-### v1.3.16
-- **`stop talking` command** - Stop speech playback
-- **dotool support** - Alternative input injection for Linux Wayland
-- **Python detection** - Better cross-platform support
-
-### v1.3.9
-- **Pure Node.js model downloads** - No system dependencies (curl/wget not required)
-
-### v1.3.3
-- **Linux support** - Full Linux platform support added
-
-## Contributing
-
-Contributions are welcome.
-
-```bash
-git clone https://github.com/Menesahin/claude-voice-extension.git
-cd claude-voice-extension
-npm install
-npm run dev
-```
-
-**Guidelines:**
-- Run `npm run lint` before committing
-- Add tests for new features
-- Follow existing code patterns
+**Wake word not detecting?** Run `claude-voice openwakeword --install` for better accuracy.
 
 ## License
 
-**PolyForm Noncommercial License 1.0.0**
-
-Free for personal use, research, education, and non-profit organizations. Commercial use requires a separate license. See [LICENSE](LICENSE) for details.
+[PolyForm Noncommercial 1.0.0](LICENSE) - Free for personal use, research, and education.
 
 ---
 
-[Documentation](https://github.com/Menesahin/claude-voice-extension#readme) |
-[Issues](https://github.com/Menesahin/claude-voice-extension/issues) |
-[Releases](https://github.com/Menesahin/claude-voice-extension/releases)
+[Issues](https://github.com/Menesahin/claude-voice-extension/issues) | [Releases](https://github.com/Menesahin/claude-voice-extension/releases)

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "claude-voice",
-  "version": "1.5.0",
+  "version": "1.5.2",
   "description": "Voice interface extension for Claude Code - TTS, STT, and wake word detection",
   "main": "dist/index.js",
   "types": "dist/index.d.ts",

package/plugin/plugin.json

@@ -1,5 +1,5 @@
 {
   "name": "claude-voice",
-  "version": "1.1.1",
+  "version": "1.5.0",
   "description": "Voice interface for Claude Code - TTS, STT, wake word detection"
 }

package/plugin/skills/listen/SKILL.md

@@ -0,0 +1,22 @@
+---
+name: listen
+description: Use when the user wants to speak a voice command. Trigger words include listen, dinle, voice, sesli komut, mikrofon, microphone, speak, konuş, söyle, ses, record, kayıt, "I want to say something", "bir şey söylemek istiyorum"
+---
+
+# Voice Listen - Activate Microphone
+
+When this skill is triggered, immediately run the following command to activate voice listening:
+
+```bash
+claude-voice listen
+```
+
+This activates the microphone and waits for the user to speak a command. The voice extension will:
+1. Play a "ping" sound to indicate it's listening
+2. Record the user's speech
+3. Transcribe it using STT (speech-to-text)
+4. Inject the transcribed text into the terminal
+
+After running the command, tell the user you're listening and they can speak their command now.
+
+If the command fails with "Daemon not running", run `claude-voice start` first, then retry `claude-voice listen`.

package/plugin/skills/voice-control/SKILL.md

@@ -1,174 +1,47 @@
 ---
 name: claude-voice
 description: Use when user asks about voice commands, TTS, STT, wake word, speech, microphone, audio, jarvis, or claude-voice CLI commands
-version: 1.1.1
+version: 1.5.0
 ---
 
 # Claude Voice Extension
 
-Voice interface for Claude Code - TTS (text-to-speech), STT (speech-to-text), and wake word detection.
+Voice interface for Claude Code - TTS, STT, and wake word detection.
 
-## Quick Start
+## Key Commands
 
 ```bash
-claude-voice start      # Start voice daemon
-claude-voice stop       # Stop daemon
-claude-voice status     # Check status
-claude-voice setup      # Interactive setup
+claude-voice start / stop / restart / status   # Daemon management
+claude-voice setup                             # Interactive setup wizard
+claude-voice doctor                            # Diagnose issues
+claude-voice listen                            # Listen for voice command (no wake word needed)
+claude-voice openai                            # Use OpenAI TTS + STT (cloud)
+claude-voice local --download                  # Use Piper TTS + Sherpa-ONNX STT (offline)
+claude-voice openwakeword --install            # Better wake word detection
+claude-voice config set <key>=<value>          # Change settings
+claude-voice model list / download <id>        # STT models
+claude-voice voice list / download <id>        # Piper TTS voices
+claude-voice test-tts "Hello"                  # Test TTS
+claude-voice logs -f                           # Follow daemon logs
 ```
 
-## CLI Commands Reference
-
-### Daemon Management
-```bash
-claude-voice start [-f]       # Start daemon (-f for foreground)
-claude-voice stop             # Stop daemon
-claude-voice restart          # Restart daemon
-claude-voice status           # Check daemon status and providers
-```
-
-### Setup & Diagnostics
-```bash
-claude-voice setup            # Interactive setup wizard
-claude-voice doctor           # Full system diagnostics
-```
-
-### Configuration
-```bash
-claude-voice config                    # View full configuration
-claude-voice config get <key>          # Get value (e.g., tts.provider)
-claude-voice config set <key>=<value>  # Set value (e.g., stt.language=tr)
-claude-voice config reset              # Reset to defaults
-claude-voice config edit               # Open in $EDITOR
-```
-
-### Hooks
-```bash
-claude-voice hooks install    # Install Claude Code hooks
-claude-voice hooks uninstall  # Remove hooks
-claude-voice hooks status     # Check hooks installation
-```
-
-### Voice Output (TTS-friendly formatting)
-```bash
-claude-voice output enable    # Enable <!-- TTS --> abstract extraction
-claude-voice output disable   # Disable voice output formatting
-claude-voice output status    # Show current settings
-claude-voice output config    # Configure (--length, --marker)
-```
-
-### STT Models (Sherpa-ONNX)
-```bash
-claude-voice model list              # List available/installed models
-claude-voice model download <id>     # Download model (whisper-tiny/base/small)
-claude-voice model remove <id>       # Remove installed model
-```
-
-### TTS Voices (Piper)
-```bash
-claude-voice voice list              # List available/installed voices
-claude-voice voice download <id>     # Download voice
-claude-voice voice remove <id>       # Remove installed voice
-claude-voice voice status            # Show current voice info
-```
-
-### Testing
-```bash
-claude-voice test-tts [text]         # Test TTS with optional text
-claude-voice test-stt <file>         # Test STT transcription
-```
-
-### Utilities
-```bash
-claude-voice voices           # List system TTS voices (macOS say)
-claude-voice devices          # List audio input devices
-claude-voice logs             # View daemon logs
-claude-voice logs -f          # Follow logs (tail -f)
-```
-
-## Configuration Options
+## Configuration
 
 Config file: `~/.claude-voice/config.json`
 
-### TTS Settings
-```
-tts.provider          # piper | macos-say | openai | elevenlabs | espeak | disabled
-tts.autoSpeak         # true - auto-speak Claude responses
-tts.maxSpeechLength   # 1500 - max chars to speak
-tts.piper.voice       # en_US-joe-medium
-tts.openai.voice      # nova | alloy | echo | fable | onyx | shimmer
-tts.macos.voice       # Samantha
-tts.macos.rate        # 200 (words per minute)
-```
-
-### STT Settings
-```
-stt.provider          # sherpa-onnx | openai | whisper-local | disabled
-stt.language          # en (ISO code: en, tr, de, fr, es, etc.)
-stt.sherpaOnnx.model  # whisper-tiny | whisper-base | whisper-small
-```
-
-### Wake Word Settings
-```
-wakeWord.enabled      # true
-wakeWord.keyword      # jarvis
-wakeWord.sensitivity  # 0.5 (0.0-1.0)
-wakeWord.playSound    # true - play Ping/Pop sounds
-```
-
-### Voice Output Settings
-```
-voiceOutput.enabled        # true - TTS-friendly abstracts
-voiceOutput.abstractMarker # <!-- TTS -->
-voiceOutput.maxAbstractLength # 200
-```
-
-### Tool TTS Settings
-```
-toolTTS.enabled       # true - announce tool completions
-toolTTS.mode          # summarize | completion
-toolTTS.tools.*       # per-tool toggle (Read, Grep, Bash, Write, Edit, etc.)
-```
+| Setting | Default | Options |
+|---------|---------|---------|
+| `tts.provider` | `macos-say` | macos-say, piper, openai, elevenlabs, espeak |
+| `stt.provider` | `sherpa-onnx` | sherpa-onnx, openai, whisper-local |
+| `stt.language` | `en` | Any ISO code (en, tr, de, fr, es...) |
+| `wakeWord.provider` | `openwakeword` | openwakeword, sherpa-onnx, picovoice |
+| `wakeWord.sensitivity` | `0.5` | 0.0 - 1.0 |
+| `voiceOutput.enabled` | `false` | true/false |
+| `toolTTS.enabled` | `false` | true/false |
 
 ## File Locations
 
-| Path | Purpose |
-|------|---------|
-| `~/.claude-voice/config.json` | User configuration |
-| `~/.claude-voice/.env` | API keys (OPENAI_API_KEY, ELEVENLABS_API_KEY) |
-| `~/.claude-voice/models/` | STT models (Whisper ONNX) |
-| `~/.claude-voice/voices/` | Piper TTS voices |
-| `~/.claude-voice/daemon.log` | Daemon logs |
-
-## Platform Support
-
-| Feature | macOS | Linux |
-|---------|-------|-------|
-| TTS | say, piper, openai | espeak, piper, openai |
-| STT | sherpa-onnx, openai | sherpa-onnx, openai |
-| Wake word | sox (rec) | arecord |
-| Terminal inject | AppleScript | xdotool (limited) |
-
-## Common Tasks
-
-**Enable voice responses:**
-```bash
-claude-voice config set tts.autoSpeak=true
-claude-voice config set voiceOutput.enabled=true
-```
-
-**Change TTS provider:**
-```bash
-claude-voice config set tts.provider=openai
-claude-voice config set tts.openai.voice=nova
-```
-
-**Change STT language:**
-```bash
-claude-voice config set stt.language=tr
-```
-
-**Download better STT model:**
-```bash
-claude-voice model download whisper-small
-```
+- Config: `~/.claude-voice/config.json`
+- API keys: `~/.claude-voice/.env`
+- Models: `~/.claude-voice/models/`
+- Logs: `~/.claude-voice/daemon.log`

package/scripts/install-plugin.js

@@ -20,19 +20,23 @@ function installPlugin(sourceDir) {
   const pluginsDir = path.join(os.homedir(), '.claude', 'plugins');
   const pluginDir = path.join(pluginsDir, PLUGIN_NAME);
   const manifestDir = path.join(pluginDir, '.claude-plugin');
-  const skillDir = path.join(pluginDir, 'skills', 'voice-control');
+  const voiceControlSkillDir = path.join(pluginDir, 'skills', 'voice-control');
+  const listenSkillDir = path.join(pluginDir, 'skills', 'listen');
 
   // Create directories
   fs.mkdirSync(manifestDir, { recursive: true });
-  fs.mkdirSync(skillDir, { recursive: true });
+  fs.mkdirSync(voiceControlSkillDir, { recursive: true });
+  fs.mkdirSync(listenSkillDir, { recursive: true });
 
   // Source paths
   const sourcePluginJson = path.join(sourceDir, 'plugin', 'plugin.json');
-  const sourceSkillMd = path.join(sourceDir, 'plugin', 'skills', 'voice-control', 'SKILL.md');
+  const sourceVoiceControlSkill = path.join(sourceDir, 'plugin', 'skills', 'voice-control', 'SKILL.md');
+  const sourceListenSkill = path.join(sourceDir, 'plugin', 'skills', 'listen', 'SKILL.md');
 
   // Destination paths
   const destPluginJson = path.join(manifestDir, 'plugin.json');
-  const destSkillMd = path.join(skillDir, 'SKILL.md');
+  const destVoiceControlSkill = path.join(voiceControlSkillDir, 'SKILL.md');
+  const destListenSkill = path.join(listenSkillDir, 'SKILL.md');
 
   // Copy plugin.json
   if (fs.existsSync(sourcePluginJson)) {
@@ -41,15 +45,25 @@ function installPlugin(sourceDir) {
     // Create minimal manifest if source doesn't exist
     const manifest = {
       name: PLUGIN_NAME,
-      version: '1.1.1',
+      version: '1.5.0',
       description: 'Voice interface for Claude Code - TTS, STT, wake word detection'
     };
     fs.writeFileSync(destPluginJson, JSON.stringify(manifest, null, 2));
   }
 
-  // Copy SKILL.md
-  if (fs.existsSync(sourceSkillMd)) {
-    fs.copyFileSync(sourceSkillMd, destSkillMd);
+  // Copy skills to plugin dir (namespaced: /claude-voice:skill-name)
+  if (fs.existsSync(sourceVoiceControlSkill)) {
+    fs.copyFileSync(sourceVoiceControlSkill, destVoiceControlSkill);
+  }
+  if (fs.existsSync(sourceListenSkill)) {
+    fs.copyFileSync(sourceListenSkill, destListenSkill);
+  }
+
+  // Also install listen skill to ~/.claude/skills/ for direct /listen invocation
+  const globalSkillsDir = path.join(os.homedir(), '.claude', 'skills', 'listen');
+  fs.mkdirSync(globalSkillsDir, { recursive: true });
+  if (fs.existsSync(sourceListenSkill)) {
+    fs.copyFileSync(sourceListenSkill, path.join(globalSkillsDir, 'SKILL.md'));
   }
 
   return pluginDir;
@@ -60,13 +74,20 @@ function installPlugin(sourceDir) {
  */
 function uninstallPlugin() {
   const pluginDir = path.join(os.homedir(), '.claude', 'plugins', PLUGIN_NAME);
+  const globalListenSkill = path.join(os.homedir(), '.claude', 'skills', 'listen');
+  let removed = false;
 
   if (fs.existsSync(pluginDir)) {
     fs.rmSync(pluginDir, { recursive: true, force: true });
-    return true;
+    removed = true;
+  }
+
+  if (fs.existsSync(globalListenSkill)) {
+    fs.rmSync(globalListenSkill, { recursive: true, force: true });
+    removed = true;
   }
 
-  return false;
+  return removed;
 }
 
 /**