fix(release): stop shipping native modules built for the wrong runtime

The released installer could not start: it carried a better-sqlite3 build for the
host Node runtime instead of Electron, so the app died immediately with a module
version mismatch when it opened its database.

Packaging now proves the Electron build of every runtime-sensitive native module
before an installer or archive exists, and installers are produced only from that
verified tree, so the mistake cannot pass silently. The release pipelines run the
same check.

The default local model also pointed at a retired model: a *.gguf name that
Ollama cannot serve, while the settings, onboarding, and guide screens
recommended an older model. All of them now use the model the service code
already preferred.

apps/desktop/src/main/services/ConfigService.ts

@@ -42,7 +42,7 @@ const CONFIG_DEFAULTS: AppConfig = {
   ttsSpeed: 1.0,
   onlineApiUrl: 'http://127.0.0.1:5000',
   localModelsDir: '',
-  llmModelId: 'gemma-2-2b-it.Q4_K_M.gguf',
+  llmModelId: 'gemma4:e4b',
   ollamaServerUrl: 'http://127.0.0.1:11434',
   appUsageMode: null,
   authToken: null,

apps/desktop/src/main/services/LocalLLMService.ts

@@ -663,7 +663,7 @@ class LocalLLMService extends EventEmitter {
     return {
       connectionState,
       serverUrl: getOllamaServerUrl(),
-      activeModel: configGet('llmModelId') || 'gemma2:2b',
+      activeModel: configGet('llmModelId') || 'gemma4:e4b',
       serverVersion: this._serverVersion
     }
   }
@@ -686,7 +686,7 @@ class LocalLLMService extends EventEmitter {
     }
 
     const serverUrl = getOllamaServerUrl()
-    const model = options?.model ?? configGet('llmModelId') ?? 'gemma2:2b'
+    const model = options?.model ?? configGet('llmModelId') ?? 'gemma4:e4b'
 
     this._abortController = new AbortController()
     this._state = LLMState.Generating

apps/desktop/src/renderer/components/OllamaGuideModal.tsx

@@ -146,7 +146,7 @@ export function OllamaGuideModal({ open, onClose }: OllamaGuideModalProps): Reac
   const { t } = useI18n()
   const [llmStatus, setLlmStatus] = useState<LLMStatus | null>(null)
   const [installedModels, setInstalledModels] = useState<LLMModel[]>([])
-  const [activeModel, setActiveModel] = useState<string>('gemma2:2b')
+  const [activeModel, setActiveModel] = useState<string>('gemma4:e4b')
   const [checking, setChecking] = useState(false)
   const [starting, setStarting] = useState(false)
   const [startMessage, setStartMessage] = useState<string | null>(null)
@@ -582,7 +582,7 @@ export function OllamaGuideModal({ open, onClose }: OllamaGuideModalProps): Reac
             <Typography sx={{ fontSize: '11px', color: d3roPalette.text.dimLabel }}>
               직접 터미널에서 다운로드하려면 아래 명령어를 실행하세요:
             </Typography>
-            <CodeBlock code="ollama pull gemma2:2b" />
+            <CodeBlock code="ollama pull gemma4:e4b" />
           </Box>
         </Box>
 

apps/desktop/src/renderer/components/OnboardingModal.tsx

@@ -49,7 +49,7 @@ export function OnboardingModal({ open, onClose }: OnboardingModalProps): React.
   // Local Ollama State
   const [ollamaStatus, setOllamaStatus] = useState<LLMStatus | null>(null)
   const [installedModels, setInstalledModels] = useState<LLMModel[]>([])
-  const [activeModel, setActiveModel] = useState<string>('gemma2:2b')
+  const [activeModel, setActiveModel] = useState<string>('gemma4:e4b')
   const [checkingOllama, setCheckingOllama] = useState(false)
   const [startingOllama, setStartingOllama] = useState(false)
   const [ollamaMsg, setOllamaMsg] = useState<string | null>(null)
@@ -498,7 +498,7 @@ export function OnboardingModal({ open, onClose }: OnboardingModalProps): React.
                   <Button
                     variant="contained"
                     startIcon={<Download size={14} />}
-                    onClick={() => handlePullModel('gemma2:2b')}
+                    onClick={() => handlePullModel('gemma4:e4b')}
                     disabled={!isOllamaConnected}
                     sx={{
                       fontWeight: 500,

apps/desktop/src/renderer/components/SettingsModal.tsx

@@ -723,7 +723,7 @@ export function SettingsModal({ open, initialTab = 0, onClose }: SettingsModalPr
                             : 'Ollama 오프라인 (서버 실행 필요)'}
                         </Typography>
                         <Typography sx={{ fontSize: '11px', color: d3roPalette.text.secondary }}>
-                          설치된 모델: {ollamaModels.length}개 | 현재 활성: {config.llmModelId ?? 'gemma2:2b'}
+                          설치된 모델: {ollamaModels.length}개 | 현재 활성: {config.llmModelId ?? 'gemma4:e4b'}
                         </Typography>
                       </Box>
                     </Box>
@@ -800,7 +800,7 @@ export function SettingsModal({ open, initialTab = 0, onClose }: SettingsModalPr
                       <InputLabel>활성 Ollama 모델</InputLabel>
                       <Select
                         label="활성 Ollama 모델"
-                        value={config.llmModelId ?? 'gemma2:2b'}
+                        value={config.llmModelId ?? 'gemma4:e4b'}
                         onChange={(e) => {
                           const val = e.target.value
                           updateConfig('llmModelId', val)
@@ -815,7 +815,7 @@ export function SettingsModal({ open, initialTab = 0, onClose }: SettingsModalPr
                         ))}
                         {/* 설치된 모델 목록이 없거나 기본 추천 추가 */}
                         {ollamaModels.length === 0 && (
-                          <MenuItem value="gemma2:2b">gemma2:2b (추천)</MenuItem>
+                          <MenuItem value="gemma4:e4b">gemma4:e4b (추천)</MenuItem>
                         )}
                         {!ollamaModels.some((m) => m.name === 'llama3.2:3b') && (
                           <MenuItem value="llama3.2:3b">llama3.2:3b</MenuItem>