fix: prevent duplicated codebase indexing on VSCode restart

src/services/code-index/__tests__/manager.spec.ts

@@ -249,7 +249,7 @@ describe("CodeIndexManager - handleSettingsChange regression", () => {
 		beforeEach(() => {
 			// Mock service factory objects
 			mockEmbedder = { embedderInfo: { name: "openai" } }
-			mockVectorStore = {}
+			mockVectorStore = { hasData: vi.fn().mockResolvedValue(false) }
 			mockScanner = {}
 			mockFileWatcher = {
 				onDidStartBatchProcessing: vi.fn(),

src/services/code-index/interfaces/vector-store.ts

@@ -62,6 +62,12 @@ export interface IVectorStore {
 	 * @returns Promise resolving to boolean indicating if the collection exists
 	 */
 	collectionExists(): Promise<boolean>
+
+	/**
+	 * Checks if the collection has any data (points)
+	 * @returns Promise resolving to boolean indicating if the collection has data
+	 */
+	hasData(): Promise<boolean>
 }
 
 export interface VectorStoreSearchResult {

src/services/code-index/manager.ts

@@ -142,14 +142,15 @@ export class CodeIndexManager {
 		}
 
 		// 5. Handle Indexing Start/Restart
-		// The enhanced vectorStore.initialize() in startIndexing() now handles dimension changes automatically
-		// by detecting incompatible collections and recreating them, so we rely on that for dimension changes
-		const shouldStartOrRestartIndexing =
-			requiresRestart ||
-			(needsServiceRecreation && (!this._orchestrator || this._orchestrator.state !== "Indexing"))
-
-		if (shouldStartOrRestartIndexing) {
-			this._orchestrator?.startIndexing() // This method is async, but we don't await it here
+		// Only start indexing if:
+		// - Configuration requires restart (settings changed)
+		// - Or this is the first initialization (no orchestrator exists yet)
+		// The orchestrator's startIndexing method will check if an existing index exists
+		// and skip the full scan if data is already present
+		const shouldStartIndexing = requiresRestart || !this._orchestrator
+
+		if (shouldStartIndexing && this._orchestrator) {
+			this._orchestrator.startIndexing() // This method is async, but we don't await it here
 		}
 
 		return { requiresRestart }

src/services/code-index/orchestrator.ts

@@ -122,6 +122,17 @@ export class CodeIndexOrchestrator {
 				await this.cacheManager.clearCacheFile()
 			}
 
+			// Check if we have an existing index with data
+			const hasExistingData = !collectionCreated && (await this.vectorStore.hasData())
+
+			if (hasExistingData) {
+				console.log("[CodeIndexOrchestrator] Existing index found with data. Skipping initial scan.")
+				// Start the file watcher without doing a full scan
+				await this._startWatcher()
+				this.stateManager.setSystemState("Indexed", "Using existing index. File watcher started.")
+				return
+			}
+
 			this.stateManager.setSystemState("Indexing", "Services ready. Starting workspace scan...")
 
 			let cumulativeBlocksIndexed = 0

src/services/code-index/processors/__tests__/file-watcher.spec.ts

@@ -118,6 +118,7 @@ describe("FileWatcher", () => {
 			upsertPoints: vi.fn().mockResolvedValue(undefined),
 			deletePointsByFilePath: vi.fn().mockResolvedValue(undefined),
 			deletePointsByMultipleFilePaths: vi.fn().mockResolvedValue(undefined),
+			hasData: vi.fn().mockResolvedValue(false),
 		}
 
 		mockIgnoreInstance = {

src/services/code-index/processors/__tests__/scanner.spec.ts

@@ -88,6 +88,7 @@ describe("DirectoryScanner", () => {
 			clearCollection: vi.fn().mockResolvedValue(undefined),
 			deleteCollection: vi.fn().mockResolvedValue(undefined),
 			collectionExists: vi.fn().mockResolvedValue(true),
+			hasData: vi.fn().mockResolvedValue(false),
 		}
 		mockCodeParser = {
 			parseFile: vi.fn().mockResolvedValue([]),

src/services/code-index/vector-store/qdrant-client.ts

@@ -488,4 +488,26 @@ export class QdrantVectorStore implements IVectorStore {
 		const collectionInfo = await this.getCollectionInfo()
 		return collectionInfo !== null
 	}
+
+	/**
+	 * Checks if the collection has any data (points)
+	 * @returns Promise resolving to boolean indicating if the collection has data
+	 */
+	async hasData(): Promise<boolean> {
+		try {
+			const collectionInfo = await this.getCollectionInfo()
+			if (!collectionInfo) {
+				return false
+			}
+
+			// Check if the collection has any points
+			// The collection info includes points_count or vectors_count depending on the version
+			const pointsCount = (collectionInfo as any).points_count || (collectionInfo as any).vectors_count || 0
+
+			return pointsCount > 0
+		} catch (error) {
+			console.warn(`[QdrantVectorStore] Error checking if collection has data:`, error)
+			return false
+		}
+	}
 }