create-byan-agent 2.1.1 → 2.2.2

package/install/package.json

@@ -1,7 +1,7 @@
 {
   "name": "create-byan-agent",
-  "version": "2.0.2",
-  "description": "Create custom AI agents in 15 minutes through intelligent interviews. GitHub Copilot CLI, VSCode, Claude Code compatible. 64 quality mantras applied automatically.",
+  "version": "2.2.2",
+  "description": "BYAN v2.2.2 - Intelligent AI agent installer with multi-platform native support (GitHub Copilot CLI, Claude Code, Codex/OpenCode)",
   "bin": {
     "create-byan-agent": "bin/create-byan-agent-v2.js"
   },

package/install/setup-turbo-whisper.js

@@ -0,0 +1,687 @@
+#!/usr/bin/env node
+
+/**
+ * Turbo Whisper Setup Script
+ * Installs Turbo Whisper voice dictation with local Whisper server
+ * 
+ * Options:
+ * - Local (CPU): Python + faster-whisper locally
+ * - Docker (GPU): Docker container with GPU support
+ * - Skip: Don't install Turbo Whisper
+ */
+
+const fs = require('fs-extra');
+const path = require('path');
+const { execSync } = require('child_process');
+const chalk = require('chalk');
+const ora = require('ora');
+
+class TurboWhisperInstaller {
+  constructor(projectRoot, mode = 'local') {
+    this.projectRoot = projectRoot;
+    this.mode = mode; // 'local', 'docker', or 'skip'
+    this.turboDir = path.join(projectRoot, '.turbo-whisper');
+    this.scriptsDir = path.join(projectRoot, 'scripts');
+    this.gpuInfo = null; // Will be populated by detectGPU()
+  }
+
+  /**
+   * Detect GPU and determine optimal Whisper model
+   * Based on official specs: https://github.com/knowall-ai/turbo-whisper
+   * @returns {Object} { hasGPU, vram, gpuName, model, modelSize }
+   */
+  detectGPU() {
+    try {
+      const result = execSync('nvidia-smi --query-gpu=name,memory.total --format=csv,noheader', { 
+        encoding: 'utf-8',
+        stdio: ['pipe', 'pipe', 'ignore']
+      }).trim();
+      
+      if (!result || result === 'NO_GPU') {
+        return { hasGPU: false, model: 'base', modelSize: '~1 GB VRAM' };
+      }
+
+      const [gpuName, vramStr] = result.split(',').map(s => s.trim());
+      const vram = parseInt(vramStr);
+
+      // Map VRAM to optimal model (official specs from GitHub)
+      // tiny: ~1 GB VRAM, base: ~1 GB, small: ~2 GB, medium: ~5 GB, large-v3: ~10 GB
+      let model, modelSize;
+      if (vram < 2000) {
+        model = 'tiny';
+        modelSize = '~1 GB VRAM';
+      } else if (vram < 4000) {
+        model = 'small';
+        modelSize = '~2 GB VRAM';
+      } else if (vram < 6000) {
+        model = 'medium';
+        modelSize = '~5 GB VRAM';
+      } else if (vram < 10000) {
+        model = 'large-v2';
+        modelSize = '~8 GB VRAM';
+      } else {
+        model = 'large-v3';
+        modelSize = '~10 GB VRAM';
+      }
+
+      return {
+        hasGPU: true,
+        vram,
+        gpuName,
+        model,
+        modelSize
+      };
+    } catch (error) {
+      return { hasGPU: false, model: 'base', modelSize: '~1 GB VRAM' };
+    }
+  }
+
+  async install() {
+    if (this.mode === 'skip') {
+      console.log(chalk.gray('Turbo Whisper installation skipped'));
+      return { success: true, skipped: true };
+    }
+
+    console.log(chalk.blue('\n📦 Installing Turbo Whisper...'));
+    console.log(chalk.gray(`Mode: ${this.mode}\n`));
+
+    try {
+      // Detect GPU first if Docker mode
+      if (this.mode === 'docker') {
+        this.gpuInfo = this.detectGPU();
+        
+        if (this.gpuInfo.hasGPU) {
+          console.log(chalk.green(`✓ GPU detected: ${this.gpuInfo.gpuName}`));
+          console.log(chalk.gray(`  VRAM: ${this.gpuInfo.vram} MB`));
+          console.log(chalk.cyan(`  Optimal model: ${this.gpuInfo.model} (${this.gpuInfo.modelSize})`));
+        } else {
+          console.log(chalk.yellow('⚠ No GPU detected, using base model (CPU)'));
+          console.log(chalk.gray('  Tip: Install nvidia drivers for GPU acceleration\n'));
+        }
+      }
+
+      await this.checkDependencies();
+      
+      if (this.mode === 'local') {
+        await this.installLocal();
+      } else if (this.mode === 'docker') {
+        await this.installDocker();
+      }
+
+      await this.createLaunchScript();
+      await this.createDocumentation();
+
+      console.log(chalk.green('\n✓ Turbo Whisper installed successfully!\n'));
+      this.printUsageInstructions();
+
+      return { success: true, mode: this.mode };
+    } catch (error) {
+      console.error(chalk.red('\n✗ Turbo Whisper installation failed:'), error.message);
+      return { success: false, error: error.message };
+    }
+  }
+
+  async checkDependencies() {
+    const spinner = ora('Checking dependencies...').start();
+
+    const required = {
+      python3: this.commandExists('python3'),
+      git: this.commandExists('git'),
+      wlCopy: this.commandExists('wl-copy'),
+      xdotool: this.commandExists('xdotool')
+    };
+
+    if (this.mode === 'docker') {
+      required.docker = this.commandExists('docker');
+    }
+
+    const missing = Object.entries(required)
+      .filter(([_, exists]) => !exists)
+      .map(([name]) => name);
+
+    if (missing.length > 0) {
+      spinner.fail('Missing dependencies');
+      console.log(chalk.yellow('\nMissing dependencies:'));
+      missing.forEach(dep => console.log(chalk.yellow(`  - ${dep}`)));
+      
+      console.log(chalk.gray('\nInstall with:'));
+      if (missing.includes('wlCopy') || missing.includes('xdotool')) {
+        console.log(chalk.gray('  sudo pacman -S wl-clipboard xdotool'));
+      }
+      if (missing.includes('docker')) {
+        console.log(chalk.gray('  sudo pacman -S docker'));
+      }
+      
+      throw new Error('Missing required dependencies');
+    }
+
+    spinner.succeed('Dependencies checked');
+  }
+
+  commandExists(command) {
+    try {
+      execSync(`which ${command}`, { stdio: 'pipe' });
+      return true;
+    } catch {
+      return false;
+    }
+  }
+
+  async installLocal() {
+    const spinner = ora('Installing Turbo Whisper (local mode)...').start();
+
+    try {
+      // Clone Turbo Whisper repository
+      const repoUrl = 'https://github.com/knowall-ai/turbo-whisper.git';
+      const installDir = path.join(process.env.HOME, '.local', 'share', 'turbo-whisper');
+
+      if (await fs.pathExists(installDir)) {
+        spinner.text = 'Turbo Whisper already installed, updating...';
+        execSync('git pull', { cwd: installDir, stdio: 'pipe' });
+      } else {
+        spinner.text = 'Cloning Turbo Whisper...';
+        execSync(`git clone ${repoUrl} ${installDir}`, { stdio: 'pipe' });
+      }
+
+      // Apply UTF-8 fixes
+      spinner.text = 'Applying UTF-8 fixes...';
+      await this.applyUTF8Fixes(installDir);
+
+      // Install Python dependencies
+      spinner.text = 'Installing Python dependencies...';
+      execSync('python3 -m venv .venv', { cwd: installDir, stdio: 'pipe' });
+      execSync('.venv/bin/pip install -e .', { cwd: installDir, stdio: 'pipe' });
+
+      // Install faster-whisper-server
+      spinner.text = 'Installing faster-whisper-server...';
+      const serverDir = path.join(process.env.HOME, 'faster-whisper-server');
+      
+      if (!(await fs.pathExists(serverDir))) {
+        execSync(`git clone https://github.com/fedirz/faster-whisper-server.git ${serverDir}`, { stdio: 'pipe' });
+        execSync('python3 -m venv .venv && .venv/bin/pip install -e .', { cwd: serverDir, stdio: 'pipe' });
+      }
+
+      spinner.succeed('Turbo Whisper installed (local mode)');
+    } catch (error) {
+      spinner.fail('Installation failed');
+      throw error;
+    }
+  }
+
+  async installDocker() {
+    const spinner = ora('Installing Turbo Whisper (Docker mode)...').start();
+
+    try {
+      // Install Turbo Whisper client
+      const installDir = path.join(process.env.HOME, '.local', 'share', 'turbo-whisper');
+      const repoUrl = 'https://github.com/knowall-ai/turbo-whisper.git';
+
+      if (await fs.pathExists(installDir)) {
+        spinner.text = 'Updating Turbo Whisper...';
+        execSync('git pull', { cwd: installDir, stdio: 'pipe' });
+      } else {
+        spinner.text = 'Cloning Turbo Whisper...';
+        execSync(`git clone ${repoUrl} ${installDir}`, { stdio: 'pipe' });
+      }
+
+      // Apply UTF-8 fixes
+      await this.applyUTF8Fixes(installDir);
+
+      // Install Python dependencies
+      spinner.text = 'Installing Python dependencies...';
+      execSync('python3 -m venv .venv && .venv/bin/pip install -e .', { cwd: installDir, stdio: 'pipe' });
+
+      // Create Docker compose file for Whisper server
+      spinner.text = 'Creating Docker configuration...';
+      await this.createDockerCompose();
+
+      spinner.succeed('Turbo Whisper installed (Docker mode)');
+      console.log(chalk.yellow('\nNote: Start Whisper server with: docker-compose up -d whisper-server'));
+    } catch (error) {
+      spinner.fail('Installation failed');
+      throw error;
+    }
+  }
+
+  async applyUTF8Fixes(installDir) {
+    // Apply UTF-8 fix to main.py
+    const mainPyPath = path.join(installDir, 'src', 'turbo_whisper', 'main.py');
+    
+    if (await fs.pathExists(mainPyPath)) {
+      let mainPy = await fs.readFile(mainPyPath, 'utf-8');
+      
+      // Check if fix already applied
+      if (!mainPy.includes('PYTHONIOENCODING')) {
+        const utf8Import = `
+import sys
+import io
+
+# Force UTF-8 encoding for all I/O operations
+if sys.platform != "win32":
+    sys.stdout = io.TextIOWrapper(sys.stdout.buffer, encoding='utf-8', errors='replace', line_buffering=True)
+    sys.stderr = io.TextIOWrapper(sys.stderr.buffer, encoding='utf-8', errors='replace', line_buffering=True)
+
+os.environ['PYTHONIOENCODING'] = 'utf-8'
+os.environ.setdefault('LC_ALL', 'fr_FR.UTF-8')
+os.environ.setdefault('LANG', 'fr_FR.UTF-8')
+`;
+        
+        // Insert after first import block
+        mainPy = mainPy.replace(
+          /import time\n/,
+          `import time\n${utf8Import}\n`
+        );
+        
+        await fs.writeFile(mainPyPath, mainPy, 'utf-8');
+      }
+    }
+
+    // Apply UTF-8 fix to typer.py
+    const typerPyPath = path.join(installDir, 'src', 'turbo_whisper', 'typer.py');
+    
+    if (await fs.pathExists(typerPyPath)) {
+      let typerPy = await fs.readFile(typerPyPath, 'utf-8');
+      
+      // Check if clipboard paste method exists
+      if (!typerPy.includes('_type_clipboard_paste')) {
+        // Add clipboard paste method implementation
+        const clipboardMethod = `
+    def _type_clipboard_paste(self, text: str) -> bool:
+        """Type text using clipboard + simulated Ctrl+Shift+V (best for UTF-8 on Wayland)."""
+        import time
+        
+        if not self.copy_to_clipboard(text):
+            print("Failed to copy to clipboard")
+            return False
+        
+        time.sleep(0.1)
+        
+        # Try Ctrl+Shift+V first (terminals on Linux)
+        if shutil.which("xdotool"):
+            try:
+                subprocess.run(
+                    ["xdotool", "key", "--clearmodifiers", "ctrl+shift+v"],
+                    check=True,
+                    capture_output=True,
+                    timeout=5
+                )
+                print("✓ Pasted via Ctrl+Shift+V")
+                return True
+            except Exception as e:
+                try:
+                    subprocess.run(
+                        ["xdotool", "key", "--clearmodifiers", "ctrl+v"],
+                        check=True,
+                        capture_output=True,
+                        timeout=5
+                    )
+                    print("✓ Pasted via Ctrl+V")
+                    return True
+                except Exception as e2:
+                    print(f"xdotool paste failed: {e}, {e2}")
+                    return False
+        
+        return False
+`;
+        
+        // Insert before _type_xdotool method
+        typerPy = typerPy.replace(
+          /    def _type_xdotool\(self, text: str\) -> bool:/,
+          `${clipboardMethod}\n    def _type_xdotool(self, text: str) -> bool:`
+        );
+        
+        await fs.writeFile(typerPyPath, typerPy, 'utf-8');
+      }
+    }
+  }
+
+  async createDockerCompose() {
+    // Use detected GPU info or fallback
+    const gpuInfo = this.gpuInfo || this.detectGPU();
+    const model = gpuInfo.model;
+    const useGPU = gpuInfo.hasGPU;
+    
+    // Format model name according to official docs:
+    // https://github.com/knowall-ai/turbo-whisper
+    const modelFullName = `Systran/faster-whisper-${model}`;
+    
+    const dockerCompose = `version: '3.8'
+
+services:
+  whisper-server:
+    image: fedirz/faster-whisper-server:latest-${useGPU ? 'cuda' : 'cpu'}
+    ports:
+      - "8000:8000"
+    environment:
+      - WHISPER__MODEL=${modelFullName}${useGPU ? `
+    deploy:
+      resources:
+        reservations:
+          devices:
+            - driver: nvidia
+              count: 1
+              capabilities: [gpu]` : ''}
+    restart: unless-stopped
+`;
+
+    const composePath = path.join(this.projectRoot, 'docker-compose.turbo-whisper.yml');
+    await fs.writeFile(composePath, dockerCompose, 'utf-8');
+    
+    // Log configuration
+    console.log(chalk.gray(`\n  Docker config: ${useGPU ? 'GPU (CUDA)' : 'CPU'} with model ${modelFullName}`));
+  }
+
+  async createLaunchScript() {
+    await fs.ensureDir(this.scriptsDir);
+
+    let launchScript;
+    
+    if (this.mode === 'local') {
+      // Mode Local: Vérifie et lance le serveur si nécessaire
+      launchScript = `#!/bin/bash
+# Launch Turbo Whisper voice dictation with automatic server startup
+
+TURBO_DIR="$HOME/.local/share/turbo-whisper"
+SERVER_DIR="$HOME/faster-whisper-server"
+SERVER_PORT=8000
+
+echo "🔍 Vérification serveur Whisper..."
+
+# Vérifier si serveur déjà en cours
+if curl -s http://localhost:$SERVER_PORT/health > /dev/null 2>&1; then
+    echo "✅ Serveur Whisper déjà actif"
+else
+    echo "⚡ Démarrage serveur Whisper..."
+    
+    # Lancer serveur en arrière-plan
+    cd "$SERVER_DIR"
+    nohup uv run uvicorn --factory faster_whisper_server.main:create_app > /tmp/whisper-server.log 2>&1 &
+    SERVER_PID=$!
+    
+    echo "⏳ Attente démarrage serveur (15 secondes)..."
+    sleep 15
+    
+    # Vérifier que le serveur répond
+    if curl -s http://localhost:$SERVER_PORT/health > /dev/null 2>&1; then
+        echo "✅ Serveur Whisper prêt (PID: $SERVER_PID)"
+    else
+        echo "❌ Erreur: Serveur n'a pas démarré"
+        echo "📋 Logs: tail -f /tmp/whisper-server.log"
+        exit 1
+    fi
+fi
+
+echo ""
+echo "🚀 Lancement Turbo Whisper..."
+echo "📍 Hotkey: Ctrl+Alt+R"
+echo "📋 Logs serveur: tail -f /tmp/whisper-server.log"
+echo ""
+
+cd "$TURBO_DIR"
+source .venv/bin/activate
+python -m turbo_whisper.main
+`;
+    } else if (this.mode === 'docker') {
+      // Get GPU info for display
+      const gpuInfo = this.gpuInfo || this.detectGPU();
+      
+      // Mode Docker: Vérifie et lance le conteneur si nécessaire
+      launchScript = `#!/bin/bash
+# Launch Turbo Whisper voice dictation with Docker server
+# Auto-detects GPU and validates configuration
+
+TURBO_DIR="$HOME/.local/share/turbo-whisper"
+SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
+PROJECT_DIR="$(dirname "$SCRIPT_DIR")"
+COMPOSE_FILE="$PROJECT_DIR/docker-compose.turbo-whisper.yml"
+SERVER_PORT=8000
+
+# Function to detect GPU
+detect_gpu() {
+    if command -v nvidia-smi &> /dev/null; then
+        GPU_INFO=$(nvidia-smi --query-gpu=name,memory.total --format=csv,noheader 2>/dev/null)
+        if [ $? -eq 0 ] && [ -n "$GPU_INFO" ]; then
+            GPU_NAME=$(echo "$GPU_INFO" | cut -d',' -f1 | xargs)
+            VRAM=$(echo "$GPU_INFO" | cut -d',' -f2 | xargs)
+            echo "✓ GPU: $GPU_NAME ($VRAM)"
+            return 0
+        fi
+    fi
+    echo "⚠ No GPU detected (running in CPU mode)"
+    return 1
+}
+
+echo "🔍 Vérification serveur Whisper Docker..."
+echo "📂 Compose file: $COMPOSE_FILE"
+echo ""
+
+# Detect GPU
+detect_gpu
+
+# Vérifier que le fichier existe
+if [ ! -f "$COMPOSE_FILE" ]; then
+    echo "❌ Erreur: $COMPOSE_FILE introuvable"
+    echo "💡 Le fichier devrait être dans le répertoire du projet"
+    exit 1
+fi
+
+# Vérifier si serveur déjà en cours
+if curl -s http://localhost:$SERVER_PORT/health > /dev/null 2>&1; then
+    echo "✅ Serveur Whisper déjà actif"
+else
+    echo "⚡ Démarrage conteneur Docker..."
+    
+    # Lancer Docker Compose
+    docker-compose -f "$COMPOSE_FILE" up -d
+    
+    echo "⏳ Attente démarrage serveur (20 secondes)..."
+    sleep 20
+    
+    # Vérifier que le serveur répond
+    if curl -s http://localhost:$SERVER_PORT/health > /dev/null 2>&1; then
+        echo "✅ Serveur Whisper prêt"
+    else
+        echo "❌ Erreur: Serveur n'a pas démarré"
+        echo "📋 Logs: docker-compose -f $COMPOSE_FILE logs"
+        exit 1
+    fi
+fi
+
+echo ""
+echo "🚀 Lancement Turbo Whisper..."
+echo "📍 Hotkey: Ctrl+Alt+R"
+echo "📋 Arrêter serveur: docker-compose -f $COMPOSE_FILE down"
+echo ""
+
+cd "$TURBO_DIR"
+source .venv/bin/activate
+python -m turbo_whisper.main
+`;
+    } else {
+      // Mode Skip (ne devrait pas arriver ici)
+      launchScript = `#!/bin/bash
+echo "⚠️  Turbo Whisper non installé"
+echo "Installez avec: npm run setup-turbo-whisper"
+exit 1
+`;
+    }
+
+    const scriptPath = path.join(this.scriptsDir, 'launch-turbo-whisper.sh');
+    await fs.writeFile(scriptPath, launchScript, 'utf-8');
+    await fs.chmod(scriptPath, '755');
+
+    // Créer script d'arrêt pour mode Docker
+    if (this.mode === 'docker') {
+      const stopScript = `#!/bin/bash
+# Stop Turbo Whisper Docker server
+
+SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
+PROJECT_DIR="$(dirname "$SCRIPT_DIR")"
+COMPOSE_FILE="$PROJECT_DIR/docker-compose.turbo-whisper.yml"
+
+if [ ! -f "$COMPOSE_FILE" ]; then
+    echo "❌ Erreur: $COMPOSE_FILE introuvable"
+    exit 1
+fi
+
+echo "🛑 Arrêt serveur Whisper Docker..."
+docker-compose -f "$COMPOSE_FILE" down
+
+echo "✅ Serveur arrêté"
+`;
+
+      const stopScriptPath = path.join(this.scriptsDir, 'stop-whisper-server.sh');
+      await fs.writeFile(stopScriptPath, stopScript, 'utf-8');
+      await fs.chmod(stopScriptPath, '755');
+    }
+
+    // Create server launch script (standalone, optionnel)
+    if (this.mode === 'local') {
+      const serverScript = `#!/bin/bash
+# Launch faster-whisper-server locally (standalone)
+
+SERVER_DIR="$HOME/faster-whisper-server"
+
+cd "$SERVER_DIR"
+echo "🚀 Starting Whisper server on http://localhost:8000"
+uv run uvicorn --factory faster_whisper_server.main:create_app
+`;
+
+
+      const serverScriptPath = path.join(this.scriptsDir, 'start-whisper-server.sh');
+      await fs.writeFile(serverScriptPath, serverScript, 'utf-8');
+      await fs.chmod(serverScriptPath, '755');
+    }
+  }
+
+  async createDocumentation() {
+    const doc = `# Turbo Whisper - Voice Dictation
+
+## Installation
+
+✅ Turbo Whisper has been installed in: \`~/.local/share/turbo-whisper\`
+${this.mode === 'local' ? '✅ Whisper server installed in: `~/faster-whisper-server`' : ''}
+${this.mode === 'docker' ? '✅ Docker configuration: `docker-compose.turbo-whisper.yml`' : ''}
+
+## Usage (Simplifié - Recommandé)
+
+### Lancement Automatique (1 commande)
+
+\`\`\`bash
+./scripts/launch-turbo-whisper.sh
+\`\`\`
+
+${this.mode === 'local' ? '**Ce script:**\n1. Vérifie si le serveur Whisper tourne\n2. Le démarre automatiquement si nécessaire (arrière-plan)\n3. Lance Turbo Whisper client\n\n**Logs serveur:** `/tmp/whisper-server.log`' : ''}
+
+${this.mode === 'docker' ? '**Ce script:**\n1. Vérifie si le conteneur Docker tourne\n2. Le démarre automatiquement si nécessaire\n3. Lance Turbo Whisper client\n\n**Arrêter serveur:** `./scripts/stop-whisper-server.sh`' : ''}
+
+## Usage Avancé (Manuel)
+
+### Démarrer Serveur Manuellement
+
+${this.mode === 'local' ? '```bash\n# Option 1: Script standalone\n./scripts/start-whisper-server.sh\n\n# Option 2: Commande directe\ncd ~/faster-whisper-server\nuv run uvicorn --factory faster_whisper_server.main:create_app\n```' : ''}
+
+${this.mode === 'docker' ? '```bash\n# Démarrer\ndocker-compose -f docker-compose.turbo-whisper.yml up -d\n\n# Vérifier\ndocker ps | grep whisper\n\n# Logs\ndocker-compose -f docker-compose.turbo-whisper.yml logs -f\n\n# Arrêter\ndocker-compose -f docker-compose.turbo-whisper.yml down\n```' : ''}
+
+### Démarrer Client Seul
+
+\`\`\`bash
+cd ~/.local/share/turbo-whisper
+source .venv/bin/activate
+python -m turbo_whisper.main
+\`\`\`
+
+### Hotkey
+
+Press **Ctrl+Alt+R** to start/stop recording.
+
+The transcribed text will be automatically typed in the active window.
+
+## Features
+
+- ✅ UTF-8 support (accents français: é, à, è, ç, â, etc.)
+- ✅ Wayland compatible
+- ✅ GPU acceleration (${this.mode === 'docker' ? 'via Docker' : 'if available'})
+- ✅ Local processing (privacy)
+- ✅ Real-time waveform visualization
+
+## Configuration
+
+Edit: \`~/.local/share/turbo-whisper/config.json\`
+
+Default Whisper server: http://localhost:8000
+
+## Troubleshooting
+
+### Caractères spéciaux ne s'affichent pas
+
+Les fixes UTF-8 ont été appliqués automatiquement. Si le problème persiste:
+
+1. Vérifiez que \`wl-clipboard\` et \`xdotool\` sont installés
+2. Redémarrez Turbo Whisper
+
+### Serveur Whisper ne démarre pas
+
+**Mode local:**
+\`\`\`bash
+cd ~/faster-whisper-server
+.venv/bin/pip install -e .
+\`\`\`
+
+**Mode Docker:**
+\`\`\`bash
+docker logs whisper-server
+\`\`\`
+
+## Documentation
+
+Voir: \`TURBO-WHISPER-INTEGRATION-COMPLETE.md\` pour détails complets.
+`;
+
+    const docPath = path.join(this.projectRoot, 'TURBO-WHISPER-SETUP.md');
+    await fs.writeFile(docPath, doc, 'utf-8');
+  }
+
+  printUsageInstructions() {
+    console.log(chalk.blue('━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━'));
+    console.log(chalk.bold('🎤 Turbo Whisper Usage:\n'));
+    
+    console.log(chalk.green('✨ Lancement Automatique (Recommandé):\n'));
+    console.log(chalk.white('   ./scripts/launch-turbo-whisper.sh'));
+    console.log(chalk.gray('   → Démarre automatiquement le serveur si nécessaire\n'));
+    
+    if (this.mode === 'local') {
+      console.log(chalk.gray('📋 Logs serveur: /tmp/whisper-server.log'));
+    } else if (this.mode === 'docker') {
+      console.log(chalk.gray('🛑 Arrêter serveur: ./scripts/stop-whisper-server.sh'));
+    }
+    
+    console.log(chalk.gray('\n🎯 Hotkey: Ctrl+Alt+R (start/stop recording)\n'));
+    
+    console.log(chalk.blue('━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━'));
+  }
+}
+
+// Export for use in main installer
+module.exports = TurboWhisperInstaller;
+
+// CLI usage
+if (require.main === module) {
+  const args = process.argv.slice(2);
+  const mode = args[0] || 'local';
+  const projectRoot = process.cwd();
+
+  const installer = new TurboWhisperInstaller(projectRoot, mode);
+  installer.install()
+    .then(result => {
+      if (result.success) {
+        process.exit(0);
+      } else {
+        process.exit(1);
+      }
+    })
+    .catch(error => {
+      console.error(chalk.red('Fatal error:'), error);
+      process.exit(1);
+    });
+}

package/install/src/byan-v2/generation/templates/default-agent.md

@@ -9,7 +9,7 @@ You must fully embody this agent's persona and follow all activation instruction
 <agent id="{{agent_id}}" name="{{agent_name}}" title="{{title}}" icon="{{icon}}">
 <activation critical="MANDATORY">
   <step n="1">Load persona from current file</step>
-  <step n="2">Load config from {project-root}/_bmad/{{module}}/config.yaml - store {user_name}, {communication_language}, {output_folder}</step>
+  <step n="2">Load config from {project-root}/_byan/{{module}}/config.yaml - store {user_name}, {communication_language}, {output_folder}</step>
   <step n="3">Show greeting using {user_name} in {communication_language}</step>
   <step n="4">Inform about `/bmad-help` command</step>
   <step n="5">WAIT for user input</step>