### Install .deb Package

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/superpowers/plans/2026-04-20-linux-appimage-auto-update.md

Installs the specified .deb package using dpkg and then launches the sokuji application. This is the initial step for users migrating from a .deb installation.

```bash
sudo dpkg -i sokuji_0.20.0-rc1_amd64.deb
sokuji
```

--------------------------------

### Start Development Server

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/superpowers/plans/2026-03-27-voxtral-asr-prototype.md

This command starts the development server. Verify that the server starts without errors and the app loads correctly. Test the keyboard shortcut (Ctrl+Shift+V) to ensure the Voxtral prototype overlay appears.

```bash
npm run dev

```

--------------------------------

### Check Installer Logs

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/build/macos-installer-guide.md

Views recent system logs for the installer process to troubleshoot installation problems.

```bash
# Check installation logs
sudo log show --predicate 'process == "installer"' --last 10m
```

--------------------------------

### Start Development Server

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/superpowers/plans/2026-03-13-unify-main-panels.md

Starts the Electron development server for the Sokuji React application.

```bash
cd /home/jiangzhuo/Desktop/kizunaai/sokuji-react && npm run electron:dev
```

--------------------------------

### Build and Start Dev Server

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/superpowers/plans/2026-05-21-supertonic-3-integration.md

Builds the application and starts the development server. This is the initial step for manual QA.

```bash
npm run build && npm run dev
```

--------------------------------

### Install Dependencies and Run Application

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/INDEX.md

Commands to install project dependencies, run the application in development mode, and build it for production.

```bash
npm ci

npm run electron:dev

npm run electron:build
```

--------------------------------

### Start Development Server

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/superpowers/plans/2026-04-18-compact-conversation-mode.md

Run this command to start the development server for manual verification. Use the web version for faster UI-only testing.

```bash
npm run electron:dev
```

```bash
npm run dev
```

--------------------------------

### Install and Test Linux Deb Package

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/superpowers/plans/2026-04-20-linux-appimage-auto-update.md

Install the generated deb package using dpkg and then launch the 'sokuji' application to ensure it installs and runs without errors. This step is recommended only on Debian-based systems.

```bash
sudo dpkg -i out/make-linux/sokuji_*_amd64.deb && sokuji
```

--------------------------------

### Install Sokuji using Homebrew

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/tutorials/macos-install.html

Use Homebrew to install Sokuji. Ensure Homebrew is installed on your system before running this command.

```bash
brew install "kizuna-ai-lab/sokuji/sokuji"
```

--------------------------------

### Install and Build Sokuji Extension

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/extension/README.md

Commands to install dependencies, build the extension for production, or build with watch mode for development.

```bash
# Install dependencies
npm install

# Build for production
npm run build

# Build for development with watch mode
npm run dev
```

--------------------------------

### Install Required Libraries

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/tutorials/linux-install.html

Install necessary libraries for Sokuji on Debian/Ubuntu-based systems.

```bash
sudo apt-get update
sudo apt-get install libgtk-3-0 libnotify4 libnss3 libxss1 \
libxtst6 xdg-utils libatspi2.0-0 libdrm2 libgbm1
```

--------------------------------

### Acquisition Lifecycle - Extension Session Start

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/superpowers/specs/2026-05-24-on-off-pipeline-gate.md

Steps for acquiring system audio on Extension when a session starts and the mode includes 'participant'.

```javascript
Extension:
  tab capture acquired by existing tabAudioRecorder.begin() flow.
```

--------------------------------

### Custom Installation Location

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/build/macos-installer-guide.md

Installs the Sokuji package to a specified custom disk volume.

```bash
# Custom installation location
installer -pkg Sokuji.pkg -target /Volumes/CustomDisk
```

--------------------------------

### Acquisition Lifecycle - Electron Session Start

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/superpowers/specs/2026-05-24-on-off-pipeline-gate.md

Steps for acquiring system audio on Electron when a session starts and the mode includes 'participant'.

```javascript
Electron:
  permissionGranted = await audioService.requestLoopbackAudioStream()
  if (!permissionGranted) → surface error in LogsPanel; skip participant
  await audioService.connectSystemAudioSource('desktop-audio-loopback')
```

--------------------------------

### Manual Smoke Test - New Install

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/superpowers/specs/2026-05-28-replay-audio-storage-toggle-design.md

Verifies the initial state and behavior after a new installation with the replay toggle off.

```bash
1. New install → toggle off; no ▶ buttons; real-time audio plays; karaoke highlight animates during streaming.
```

--------------------------------

### Start Electron in Development Mode

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/superpowers/plans/2026-04-22-bing-translator-integration.md

Run this command to start the Electron application in development mode, which is the first step in the manual testing process.

```bash
npm run electron:dev
```

--------------------------------

### PostHog Installation and Initialization

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/extension/extension-popup-analytics.md

Imports and initializes the PostHog library using a locally installed package. Configuration is managed via environment variables, with a fallback to a default host. Includes a helper function to check if analytics are enabled.

```javascript
// Import PostHog from installed package
import PostHog from 'posthog-js-lite';

// Analytics configuration - set via environment or leave empty to disable
const ANALYTICS_CONFIG = {
  POSTHOG_KEY: process.env.POSTHOG_KEY || '',
  POSTHOG_HOST: process.env.POSTHOG_HOST || 'https://us.i.posthog.com'
};

// Helper to check if analytics is enabled
function isAnalyticsEnabled() {
  return Boolean(ANALYTICS_CONFIG.POSTHOG_KEY);
}
```

--------------------------------

### Build Sokuji Application and Installer

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/build/macos-installer-guide.md

Commands to build the virtual audio driver, the PKG installer, and leverage npm scripts for the Electron app packaging. These scripts automate the build and packaging process.

```bash
# Build the virtual audio driver
./build-sokuji-driver.sh

# Build PKG installer
./build-pkg.sh

# Or use npm scripts for Electron app
npm run make:pkg
```

--------------------------------

### Update Basic-Mode Start Button Tooltip

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/superpowers/plans/2026-05-22-symmetric-channel-architecture.md

Extends the existing tooltip for the basic-mode Start button to include the case where no audio channels are configured. This is used when the 'canStartSession' gate is false and 'anyChannelWillStart' is also false.

```tsx
<button
  className={`main-action-btn ${isSessionActive ? 'stop' : 'start'}`}
  onClick={isSessionActive ? disconnectConversation : connectConversation}
  disabled={!canStartSession && !isSessionActive}
  title={
    !canStartSession && !isSessionActive
      ? !anyChannelWillStart
        ? t('mainPanel.noChannelConfigured', 'Enable microphone or participant audio before starting.')
        : provider === Provider.LOCAL_INFERENCE
          ? t('mainPanel.localModelsRequired', 'Download required models in settings to start.')
          : undefined
      : undefined
  }
>

```

--------------------------------

### Asynchronous Audio Decoding Initialization

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/model-packs/asr/omnilingual.html

Initializes and executes the audio decoding process asynchronously, including setup and finalization steps.

```javascript
})();
```

--------------------------------

### Upload Unsigned Windows Installer for Signing

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/superpowers/specs/2026-03-13-signpath-windows-signing-design.md

Uploads the unsigned Windows installer artifact for subsequent signing. Use when the GitHub ref starts with 'refs/tags/v' and the OS is Windows.

```yaml
- name: Upload unsigned Windows installer for signing
  if: startsWith(github.ref, 'refs/tags/v') && matrix.os == 'windows-latest'
  uses: actions/upload-artifact@v4
  with:
    name: windows-unsigned
    path: out/make/squirrel.windows/x64/*Setup*.exe
```

--------------------------------

### Get and Sync Distinct ID in Sidepanel (React)

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/extension/extension-posthog-distinct-id.md

This example shows how to get the current PostHog distinct ID and trigger a sync to the background script using React hooks within a component. Requires access to `useAnalytics`.

```javascript
// Get distinct_id from PostHog in sidepanel (using React hook)
// Note: This should be run inside a React component that has access to useAnalytics
const { getDistinctId, syncDistinctIdToBackground } = useAnalytics();
console.log('Current PostHog distinct_id:', getDistinctId());
```

--------------------------------

### Track Extension Installation

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/app/app-analytics-events.md

Triggered when the browser extension is installed. Records the version of the installed extension.

```typescript
trackEvent('extension_installed', {
  extension_version: '1.0.0'
});
```

--------------------------------

### Sign the PKG Installer

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/build/macos-installer-guide.md

Signs the unsigned PKG installer file using a Developer ID Installer certificate. Replace 'Developer ID Installer: Your Name (XXXXXXXXXX)' with your actual certificate identity.

```bash
productsign --sign "Developer ID Installer: Your Name (XXXXXXXXXX)" \
  Sokuji-unsigned.pkg \
  Sokuji.pkg
```

--------------------------------

### Run Development Server

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/superpowers/plans/2026-05-12-tutorial-subtitle-mode.md

Start the development server to smoke-test the newly added Subtitle Mode route locally.

```bash
cd web && npm run dev
```

--------------------------------

### Install electron-updater Dependency

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/superpowers/plans/2026-03-16-auto-update.md

Installs the electron-updater package as a project dependency. Verify the installation by checking the package.json file.

```bash
cd /home/jiangzhuo/Desktop/kizunaai/sokuji-react/.claude/worktrees/auto-update-design
npm install electron-updater
```

```bash
node -e "require('electron-updater')" 2>&1 || echo "OK - electron main process module"
grep '"electron-updater"' package.json
```

```bash
git add package.json package-lock.json
git commit -m "chore: add electron-updater dependency"
```

--------------------------------

### Build the Extension

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/superpowers/specs/2026-05-13-subtitle-mode-extension-design.md

Navigate to the extension directory and build the project using npm.

```bash
cd extension && npm run build
```

--------------------------------

### Install BlackHole via Homebrew

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/resources/drivers/SokujiVirtualAudio.driver/Contents/Resources/README.md

Installs different channel versions of BlackHole using Homebrew. Ensure Homebrew is installed before running these commands.

```bash
brew install blackhole-2ch
```

```bash
brew install blackhole-16ch
```

```bash
brew install blackhole-64ch
```

--------------------------------

### Post-installation Script for Virtual Audio Driver

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/build/macos-installer-guide.md

Installs the virtual audio driver, sets permissions, and restarts CoreAudio to load the driver. Requires sudo privileges.

```bash
#!/bin/bash

# Install virtual audio driver
DRIVER_SOURCE="/Applications/Sokuji.app/Contents/Resources/drivers/SokujiVirtualAudio.driver"
DRIVER_DEST="/Library/Audio/Plug-Ins/HAL/"

if [ -d "$DRIVER_SOURCE" ]; then
    # Copy driver to system location
    sudo cp -R "$DRIVER_SOURCE" "$DRIVER_DEST"

    # Set proper permissions
    sudo chown -R root:wheel "$DRIVER_DEST/SokujiVirtualAudio.driver"
    sudo chmod -R 755 "$DRIVER_DEST/SokujiVirtualAudio.driver"

    # Restart CoreAudio to load the driver
    sudo killall coreaudiod
fi

exit 0
```

--------------------------------

### Run macOS Installer Test Script

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/build/macos-installer-guide.md

Executes the test script for the macOS installer, which handles installation, verification, audio testing, signature validation, and uninstallation.

```bash
# Run the test script
./test-macos-installer.sh
```

--------------------------------

### Install Updated Dependency

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/superpowers/plans/2026-03-27-voxtral-asr-prototype.md

Install the updated npm dependencies after modifying package.json.

```bash
npm install
```

--------------------------------

### Build Extension

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/superpowers/plans/2026-05-31-extension-unsupported-site-redesign.md

Run this command to build the extension and confirm it bundles without errors. The `extension/dist/popup.js` file should be regenerated.

```bash
npm run extension:build
```

--------------------------------

### Initialize Electron Audio Loopback

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/SYSTEM_AUDIO_CAPTURE.md

Sets up the necessary handlers for capturing system audio in an Electron application. This should be called in the main process.

```javascript
const { initMain } = require('electron-audio-loopback');
initMain();
```

--------------------------------

### Electron Forge PKG Configuration

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/build/macos-installer-guide.md

Configuration for Electron Forge to create a macOS PKG installer. Specifies the installer name, signing identity, script location, installation path, and UI files.

```javascript
{
  name: '@electron-forge/maker-pkg',
  config: {
    name: 'Sokuji',
    identity: 'Developer ID Installer: Your Name (XXXXXXXXXX)',
    scripts: 'pkg-scripts',
    installLocation: '/Applications',
    welcome: 'resources/installer-welcome.html',
    conclusion: 'resources/installer-conclusion.html'
  }
}
```

--------------------------------

### Adding a New Onboarding Step

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/guides/onboarding-guide.md

Example of how to define a new step for the onboarding tour, specifying its target element, content, title, and placement.

```typescript
{
  target: '.new-element',
  content: 'Description of the new feature.',
  title: 'Step Title',
  placement: 'bottom',
}
```

--------------------------------

### Silent Installation

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/build/macos-installer-guide.md

Performs a silent installation of the Sokuji package without user interaction.

```bash
# Silent installation
installer -pkg Sokuji.pkg -target / -verboseR
```

--------------------------------

### Run Development Servers

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/INDEX.md

Commands to start development servers for the Electron app, React app, or Chrome extension.

```bash
# Electron app
npm run electron:dev

# React app only
npm run dev

# Extension development
cd extension && npm run dev
```

--------------------------------

### Commit mpg123-decoder Installation

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/superpowers/plans/2026-04-11-edge-tts-integration.md

Add and commit the changes to package.json and package-lock.json after installing the mpg123-decoder dependency.

```bash
git add package.json package-lock.json
git commit -m "chore: add mpg123-decoder dependency for Edge TTS MP3 decoding"
```

--------------------------------

### Original Session Start Recording Logic

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/superpowers/plans/2026-04-28-push-to-translate.md

This snippet shows the original logic for determining whether to start recording at the session start, based on turn detection settings. It's used to identify the block that needs modification.

```typescript
// Start recording if using server VAD and input device is turned on
// Note: Skip manual recording for WebRTC mode - audio flows via MediaStreamTrack
let turnDetectionDisabled = false;
if (isOpenAICompatible(provider)) {
  const settings =
    provider === Provider.OPENAI ? openAISettings :
    provider === Provider.OPENAI_COMPATIBLE ? openAICompatibleSettings :
    provider === Provider.KIZUNA_AI ? kizunaAISettings :
    null;
  turnDetectionDisabled = settings ? settings.turnDetectionMode === 'Disabled' : false;
} else if (provider === Provider.VOLCENGINE_AST2) {
  turnDetectionDisabled = volcengineAST2Settings.turnDetectionMode === 'Push-to-Talk';
} else if (provider === Provider.LOCAL_INFERENCE) {
  turnDetectionDisabled = localInferenceSettings.turnDetectionMode === 'Push-to-Talk';
} else if (provider === Provider.GEMINI) {
  turnDetectionDisabled = geminiSettings.turnDetectionMode === 'Push-to-Talk';
}
...
// Check if provider uses native audio capture (OpenAI WebRTC or PalabraAI/LiveKit)
const usesNativeCapture = ClientFactory.usesNativeAudioCapture(provider, useWebRTC ? 'webrtc' : 'websocket');
...
if (!usesNativeCapture && !turnDetectionDisabled && isInputDeviceOn && audioServiceRef.current) {
  let audioCallbackCount = 0;
  await audioServiceRef.current.startRecording(selectedInputDevice?.deviceId, (data) => {
    if (clientRef.current) {
      if (audioCallbackCount % 100 === 0) {
        console.debug(`[Sokuji] [MainPanel] Sending audio to client: chunk ${audioCallbackCount}, PCM length: ${data.mono.length}`);
      }
      audioCallbackCount++;
      clientRef.current.appendInputAudio(data.mono);
    }
  });
}
```

--------------------------------

### Build and Test Project

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/superpowers/plans/2026-05-27-panel-chrome-unification.md

Run this command to typecheck, build, and test the entire project. Ensure all outputs are successful.

```bash
npx tsc --noEmit && npm run build && npm test
```

--------------------------------

### Check Installed Application Version

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/build/macos-installer-guide.md

Retrieves the version string of the installed Sokuji application from its Info.plist file.

```bash
# Check installed version
defaults read /Applications/Sokuji.app/Contents/Info.plist CFBundleShortVersionString
```

--------------------------------

### Verify Image Directory Creation

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/superpowers/plans/2026-05-12-tutorial-subtitle-mode.md

After creating the placeholder README, verify that the directory and file have been successfully created in the file system.

```bash
ls web/public/features/subtitle-mode/
```

--------------------------------

### Build Project

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/superpowers/plans/2026-04-01-granite-speech-integration.md

Run this command to ensure the project builds successfully without errors.

```bash
npm run build

```

--------------------------------

### Install .deb Package via Terminal

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/tutorials/linux-install.html

Use these commands to install the .deb package if you encounter dependency issues.

```bash
cd ~/Downloads
sudo dpkg -i sokuji_*.deb
sudo apt-get install -f
```

--------------------------------

### Verify mpg123-decoder Installation

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/superpowers/plans/2026-04-11-edge-tts-integration.md

Verify that the mpg123-decoder package has been successfully installed and is present in the project's dependency tree.

```bash
npm ls mpg123-decoder
```

--------------------------------

### Create Directory

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/superpowers/plans/2026-05-14-more-color-options.md

Use mkdir to create a new directory for the component. No output is expected if the directory is created successfully.

```bash
mkdir -p src/components/Display
```

--------------------------------

### Check Driver Installation Status

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/build/macos-installer-guide.md

Verifies if the Sokuji virtual audio driver is installed in the HAL plugins directory.

```bash
# Check if driver is installed
ls -la /Library/Audio/Plug-Ins/HAL/SokujiVirtualAudio.driver
```

--------------------------------

### Registering New Platform Plugin

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/extension/extension-gather-town-integration.md

Demonstrates how to register a new platform plugin by adding it to the global `sokujiSitePlugins` object, mapping the platform's hostname to its corresponding plugin object.

```javascript
window.sokujiSitePlugins = {
  'platform.domain.com': newPlatformPlugin,
  // ... other plugins
};
```

--------------------------------

### Make AppImage Executable and Run

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/tutorials/linux-install.html

After downloading the AppImage, make it executable and then run it from the terminal.

```bash
chmod +x sokuji-*.AppImage
./sokuji-*.AppImage
```

--------------------------------

### Build and Verify Project

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/superpowers/plans/2026-03-13-unify-main-panels.md

Builds the React project and checks the last 20 lines of output for any errors.

```bash
cd /home/jiangzhuo/Desktop/kizunaai/sokuji-react && npm run build 2>&1 | tail -20
```

--------------------------------

### Install mpg123-decoder Dependency

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/superpowers/plans/2026-04-11-edge-tts-integration.md

Install the mpg123-decoder npm package. This is the first step in enabling MP3 decoding for Edge TTS.

```bash
npm install mpg123-decoder
```

--------------------------------

### Conditional Start Recording Logic

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/superpowers/plans/2026-04-28-push-to-translate.md

This snippet modifies the `startRecording` function to conditionally execute recorder start logic based on the `currentTurnDetectionMode`. For 'Push-to-Translate' mode, it bypasses direct audio service calls and only resets state. For other modes, it proceeds with the standard recorder start procedure.

```typescript
try {
  const isPushToTranslate = currentTurnDetectionMode === 'Push-to-Translate';

  if (isPushToTranslate) {
    // Push-to-translate: recorder is already running continuously.
    // Just reset chunk counter; the unified passthrough useEffect will mute passthrough
    // (because isRecording is now true), and the gated recording callback (set up at
    // session start) will start forwarding audio to the AI client.
    pttVoiceChunkCountRef.current = 0;
    return;
  }

  // Pure PTT modes (Push-to-Talk / Disabled): start the recorder fresh on each hold.
  // Note: We no longer interrupt playing audio when recording starts
  // This allows for simultaneous recording and playback

  // Check if the recorder is in a valid state
  const recorder = audioService.getRecorder();
  if (recorder.isRecording()) {
    // If somehow we're already recording, pause first
    console.warn('[Sokuji] [MainPanel] ModernAudioRecorder was already recording, pausing first');
    await audioService.pauseRecording();
  }

  // Start recording
  pttVoiceChunkCountRef.current = 0;  // Reset non-silent chunk counter
  let pttAudioCallbackCount = 0;
  await audioService.startRecording(selectedInputDevice?.deviceId, (data) => {
    if (client) {
      // Debug logging for push-to-talk (every 50 chunks)
      if (pttAudioCallbackCount % 50 === 0) {
        console.debug(`[Sokuji] [MainPanel] PTT: Sending audio to client: chunk ${pttAudioCallbackCount}, PCM length: ${data.mono.length}`);
      }
      pttAudioCallbackCount++;

      // Track non-silent audio chunks for empty request detection
      if (!isSilentAudio(data.mono)) {
        pttVoiceChunkCountRef.current++;
      }

      client.appendInputAudio(data.mono);
    }
  });
} catch (error) {
  console.error('[Sokuji] [MainPanel] Error starting recording:', error);
  setIsRecording(false);
}
```

--------------------------------

### Build and Test Project

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/superpowers/plans/2026-05-27-panel-chrome-unification.md

Build the project and run tests to confirm the refactoring did not break functionality.

```bash
npm run build && npm run test
```

--------------------------------

### Run Existing Tests

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/superpowers/plans/2026-03-19-webgpu-shader-f16-dtype.md

Execute the existing test suite using Vitest. All tests should pass, ignoring unrelated pre-existing failures in settingsStore.

```bash
npx vitest run
```

--------------------------------

### Verify electron-builder Installation

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/superpowers/plans/2026-04-20-linux-appimage-auto-update.md

Checks if electron-builder has been installed correctly by printing its version number. This command should output a version number without errors.

```bash
npx electron-builder --version
```

--------------------------------

### Configure Provider to Doubao AST 2.0

Source: https://github.com/kizuna-ai-lab/sokuji/blob/main/docs/superpowers/plans/2026-05-14-volcengine-ast2-bidirectional-mode.md

In the application settings, navigate to the Provider section and switch to 'Doubao AST 2.0' (volcengine_ast2). Ensure a valid App Key / Access Token is configured if not already present. This step is necessary to test the Volcengine AST2 specific features.

```bash
In Settings → Provider, switch to **Doubao AST 2.0** (`volcengine_ast2`). Enter a valid App Key / Access Token if not already configured.
```