Merge pull request #369 from StroepWafel/globalSearch-improvements

Global search improvements

.npmrc

@@ -0,0 +1,2 @@
+legacy-peer-deps=true
+

package.json

@@ -5,8 +5,10 @@
   "description": "Enhance SEQTA Learn's usability and aesthetics! A fork of BetterSEQTA to continue development add add heaps more features!",
   "browserslist": "> 0.5%, last 2 versions, not dead",
   "scripts": {
+    "autoaudit": "npm audit && npm audit fix && npm run build",
     "dev": "cross-env MODE=chrome vite dev",
     "dev:firefox": "cross-env MODE=firefox vite build --watch",
+    "compile": "npm i && npm run build",
     "build": "cross-env MODE=chrome vite build && cross-env MODE=firefox vite build",
     "build:chrome": "cross-env MODE=chrome vite build",
     "build:firefox": "cross-env MODE=firefox vite build",

src/css/injected.scss

@@ -9,7 +9,7 @@
   background: var(--better-main) !important;
   --navy: #1a1a1a !important;
   --auto-background: var(--better-pale, var(--background-secondary)) !important;
-  font-family: Rubik, sans-serif !important;
+  font-family: Rubik, system-ui, -apple-system, BlinkMacSystemFont, "Segoe UI", Roboto, "Helvetica Neue", Arial, sans-serif !important;
 }
 
 ::view-transition-old(root),
@@ -36,7 +36,7 @@ body,
 .legacy-root option,
 .legacy-root .input,
 html {
-  font-family: Rubik, sans-serif !important;
+  font-family: Rubik, system-ui, -apple-system, BlinkMacSystemFont, "Segoe UI", Roboto, "Helvetica Neue", Arial, sans-serif !important;
 }
 
 select option {
@@ -56,7 +56,7 @@ select {
   background: var(--auto-background) !important;
 }
 :root * {
-  font-family: Rubik, sans-serif !important;
+  font-family: Rubik, system-ui, -apple-system, BlinkMacSystemFont, "Segoe UI", Roboto, "Helvetica Neue", Arial, sans-serif !important;
   --theme-fg-parts: white;
 }
 .extension-editor {
@@ -302,7 +302,7 @@ select {
 
       .material-icons {
         font-size: 0px !important;
-        font-family: Rubik, sans-serif !important;
+        font-family: Rubik, system-ui, -apple-system, BlinkMacSystemFont, "Segoe UI", Roboto, "Helvetica Neue", Arial, sans-serif !important;
         &::before {
           font-size: 18px !important;
           content: "Search" !important;
@@ -413,7 +413,7 @@ ul.magicDelete > li.deleting {
   background: var(--better-main) !important;
   color: var(--text-color);
   border-right: none;
-  font-family: Rubik, sans-serif !important;
+  font-family: Rubik, system-ui, -apple-system, BlinkMacSystemFont, "Segoe UI", Roboto, "Helvetica Neue", Arial, sans-serif !important;
 }
 #menu li > label > svg,
 #menu section > label > svg {

src/plugins/built-in/globalSearch/lazy.ts

@@ -42,32 +42,69 @@ const settings = defineSettings({
 
       if (confirmed) {
         try {
-          // Dynamically import the worker manager to avoid loading heavy dependencies
+          // Dynamically import modules to avoid loading heavy dependencies
           const { VectorWorkerManager } = await import("./src/indexing/worker/vectorWorkerManager");
-          const workerManager = VectorWorkerManager.getInstance();
+          const { resetDatabase } = await import("./src/indexing/db");
-          await workerManager.resetWorker();
-          console.log("Vector worker reset successfully");
-        } catch (e) {
-          console.warn("Failed to reset vector worker:", e);
-        }
           
-        // Delete both 'embeddiaDB' and 'betterseqta-index' using native IndexedDB APIs
+          // Reset vector worker first
-        const deleteDb = (dbName: string) => {
+          try {
-          return new Promise<void>((resolve, reject) => {
+            const workerManager = VectorWorkerManager.getInstance();
-            const req = indexedDB.deleteDatabase(dbName);
+            await workerManager.resetWorker();
-            req.onsuccess = () => resolve();
+            console.log("Vector worker reset successfully");
-            req.onerror = () => reject(req.error);
+          } catch (e) {
-            req.onblocked = () => {
+            console.warn("Failed to reset vector worker:", e);
-              reject(new Error(`One database is open, failed to remove: ${dbName}`));
+          }
-            };
+
-          });
+          // Close all database connections properly before deletion
-        };
+          try {
-        try {
+            await resetDatabase();
-          await deleteDb("embeddiaDB");
+            console.log("betterseqta-index database closed and reset");
-          await deleteDb("betterseqta-index");
+          } catch (e) {
-          alert("Search index and storage have been reset.");
+            console.warn("Failed to reset betterseqta-index database:", e);
+          }
+
+          // Wait a bit for connections to fully close
+          await new Promise(resolve => setTimeout(resolve, 100));
+
+          // Delete embeddiaDB (vector search database)
+          const deleteDb = (dbName: string) => {
+            return new Promise<void>((resolve, reject) => {
+              const req = indexedDB.deleteDatabase(dbName);
+              req.onsuccess = () => {
+                console.log(`Successfully deleted database: ${dbName}`);
+                resolve();
+              };
+              req.onerror = () => {
+                console.error(`Error deleting database ${dbName}:`, req.error);
+                reject(req.error);
+              };
+              req.onblocked = () => {
+                console.warn(`Database ${dbName} deletion blocked - connections still open`);
+                // Wait and retry once
+                setTimeout(() => {
+                  const retryReq = indexedDB.deleteDatabase(dbName);
+                  retryReq.onsuccess = () => {
+                    console.log(`Successfully deleted database on retry: ${dbName}`);
+                    resolve();
+                  };
+                  retryReq.onerror = () => reject(retryReq.error);
+                  retryReq.onblocked = () => {
+                    reject(new Error(`One database is open, failed to remove: ${dbName}. Please close other tabs and try again.`));
+                  };
+                }, 500);
+              };
+            });
+          };
+          
+          try {
+            await deleteDb("embeddiaDB");
+            await deleteDb("betterseqta-index");
+            alert("Search index and storage have been reset successfully.");
+          } catch (e) {
+            alert("Failed to reset one or more databases: " + String(e) + "\n\nTry closing other browser tabs and try again.");
+          }
         } catch (e) {
-          alert("Failed to reset one or more databases: " + String(e));
+          alert("Failed to reset index: " + String(e));
         }
       }
     },

src/plugins/built-in/globalSearch/src/components/SearchBar.svelte

@@ -35,6 +35,8 @@
   let isIndexing = $state(false);
   let completedJobs = $state(0);
   let totalJobs = $state(0);
+  let indexingStatus = $state<string | null>(null);
+  let indexingDetail = $state<string | null>(null);
 
   let commandPalleteOpen = $state(false);
   let searchTerm = $state('');
@@ -110,10 +112,12 @@
 
   onMount(() => {
     const progressHandler = (event: CustomEvent) => {
-      const { completed, total, indexing } = event.detail;
+      const { completed, total, indexing, status, detail } = event.detail;
       completedJobs = completed;
       totalJobs = total;
       isIndexing = indexing;
+      indexingStatus = status || null;
+      indexingDetail = detail || null;
     };
 
     window.addEventListener('indexing-progress', progressHandler as EventListener);
@@ -168,6 +172,9 @@
         term, 
         commandsFuse, 
         commandIdToItemMap,
+        dynamicContentFuse,
+        dynamicIdToItemMap,
+        true, // sortByRecent
       );
     } else {
       combinedResults = [];
@@ -176,13 +183,19 @@
     isLoading = false;
   };
 
-  const debouncedPerformSearch = debounce(performSearch, 20);
+  // Optimized debounce: shorter delay for better responsiveness
+  const debouncedPerformSearch = debounce(performSearch, 50);
 
   $effect(() => {
     if (commandPalleteOpen) {
       if (searchTerm === '') {
+        // Immediate search for empty query (shows recent items)
+        performSearch();
+      } else if (searchTerm.length <= 2) {
+        // Immediate search for very short queries
         performSearch();
       } else {
+        // Debounced search for longer queries
         debouncedPerformSearch();
       }
       tick().then(() => searchbar?.focus()); 
@@ -389,19 +402,6 @@
                 {@render Shortcut({ text: 'Select', keybind: ['↵']})}
                 {/if}
               </div>
-              {#if isIndexing}
-                <div class="inset-x-0 top-0">
-                  <div class="absolute right-2 -bottom-4 text-[10px] text-zinc-500 dark:text-zinc-400">
-                    Indexing
-                  </div>
-                  <div class="overflow-hidden h-0.5 bg-zinc-200 dark:bg-zinc-700">
-                    <div 
-                      class="h-full bg-blue-500 transition-all duration-300 ease-out"
-                      style="width: {(completedJobs / totalJobs) * 100}%"
-                    ></div>
-                  </div>
-                </div>
-              {/if}
             </div>
           {/if}
         </div>

src/plugins/built-in/globalSearch/src/core/index.ts

@@ -14,6 +14,7 @@ import { initVectorSearch } from "../search/vector/vectorSearch";
 import { cleanupSearchBar, mountSearchBar } from "./mountSearchBar";
 import { IndexedDbManager } from "embeddia";
 import { VectorWorkerManager } from "../indexing/worker/vectorWorkerManager";
+import { checkAndHandleUpdate } from "../utils/versionCheck";
 
 // Platform-aware default hotkey
 const getDefaultHotkey = () => {
@@ -50,31 +51,67 @@ const settings = defineSettings({
 
       if (confirmed) {
         try {
-          // Reset the vector worker first
+          // Import resetDatabase function to properly close connections
-          const workerManager = VectorWorkerManager.getInstance();
+          const { resetDatabase } = await import("../indexing/db");
-          await workerManager.resetWorker();
-          console.log("Vector worker reset successfully");
-        } catch (e) {
-          console.warn("Failed to reset vector worker:", e);
-        }
           
-        // Delete both 'embeddiaDB' and 'betterseqta-index' using native IndexedDB APIs
+          // Reset the vector worker first
-        const deleteDb = (dbName: string) => {
+          try {
-          return new Promise<void>((resolve, reject) => {
+            const workerManager = VectorWorkerManager.getInstance();
-            const req = indexedDB.deleteDatabase(dbName);
+            await workerManager.resetWorker();
-            req.onsuccess = () => resolve();
+            console.log("Vector worker reset successfully");
-            req.onerror = () => reject(req.error);
+          } catch (e) {
-            req.onblocked = () => {
+            console.warn("Failed to reset vector worker:", e);
-              reject(new Error(`One database is open, failed to remove: ${dbName}`));
+          }
-            };
+
-          });
+          // Close all database connections properly before deletion
-        };
+          try {
-        try {
+            await resetDatabase();
-          await deleteDb("embeddiaDB");
+          } catch (e) {
-          await deleteDb("betterseqta-index");
+            console.warn("Failed to reset betterseqta-index database:", e);
-          alert("Search index and storage have been reset.");
+          }
+
+          // Wait a bit for connections to fully close
+          await new Promise(resolve => setTimeout(resolve, 100));
+
+          // Delete embeddiaDB (vector search database)
+          const deleteDb = (dbName: string) => {
+            return new Promise<void>((resolve, reject) => {
+              const req = indexedDB.deleteDatabase(dbName);
+              req.onsuccess = () => {
+                console.log(`Successfully deleted database: ${dbName}`);
+                resolve();
+              };
+              req.onerror = () => {
+                console.error(`Error deleting database ${dbName}:`, req.error);
+                reject(req.error);
+              };
+              req.onblocked = () => {
+                console.warn(`Database ${dbName} deletion blocked - connections still open`);
+                // Wait and retry once
+                setTimeout(() => {
+                  const retryReq = indexedDB.deleteDatabase(dbName);
+                  retryReq.onsuccess = () => {
+                    console.log(`Successfully deleted database on retry: ${dbName}`);
+                    resolve();
+                  };
+                  retryReq.onerror = () => reject(retryReq.error);
+                  retryReq.onblocked = () => {
+                    reject(new Error(`One database is open, failed to remove: ${dbName}. Please close other tabs and try again.`));
+                  };
+                }, 500);
+              };
+            });
+          };
+          
+          try {
+            await deleteDb("embeddiaDB");
+            await deleteDb("betterseqta-index");
+            alert("Search index and storage have been reset successfully.");
+          } catch (e) {
+            alert("Failed to reset one or more databases: " + String(e) + "\n\nTry closing other browser tabs and try again.");
+          }
         } catch (e) {
-          alert("Failed to reset one or more databases: " + String(e));
+          alert("Failed to reset index: " + String(e));
         }
       }
     },
@@ -114,6 +151,27 @@ const globalSearchPlugin: Plugin<typeof settings> = {
   run: async (api) => {
     const appRef = { current: null };
 
+    // Check for extension updates and clear caches if needed
+    // Use a timeout to avoid blocking initialization
+    setTimeout(async () => {
+      try {
+        const wasUpdated = await checkAndHandleUpdate();
+        if (wasUpdated) {
+          console.log("[Global Search] Extension updated - caches cleared");
+        }
+      } catch (error: any) {
+        // Handle CSS preload errors and other failures gracefully
+        // These can happen in Firefox or when assets aren't available
+        if (error?.message?.includes("preload CSS") || 
+            error?.message?.includes("MIME type") || 
+            error?.message?.includes("NS_ERROR_CORRUPTED_CONTENT")) {
+          console.debug("[Global Search] Version check skipped due to asset loading restrictions:", error.message);
+        } else {
+          console.warn("[Global Search] Failed to check for updates:", error);
+        }
+      }
+    }, 100);
+
     try {
       await IndexedDbManager.create("embeddiaDB", "embeddiaObjectStore", {
         primaryKey: "id",
@@ -126,10 +184,16 @@ const globalSearchPlugin: Plugin<typeof settings> = {
 
     initVectorSearch();
 
-    // Warm up vector worker in background to improve initial response time
+    // Warm up vector worker in background to improve initial response time (skip in Firefox)
     setTimeout(async () => {
       try {
-        VectorWorkerManager.getInstance();
+        // Only initialize worker if vector search is supported
+        const { isVectorSearchSupported } = await import("../utils/browserDetection");
+        if (isVectorSearchSupported()) {
+          VectorWorkerManager.getInstance();
+        } else {
+          console.debug("[Global Search] Skipping vector worker warm-up (Firefox detected - using text search only)");
+        }
       } catch (error) {
         console.warn("[Global Search] Vector worker warm-up failed:", error);
       }

src/plugins/built-in/globalSearch/src/core/mountSearchBar.ts

@@ -8,7 +8,7 @@ import browser from "webextension-polyfill";
 export function mountSearchBar(
   titleElement: Element,
   api: any,
-  appRef: { current: any; storageChangeHandler?: any },
+  appRef: { current: any; storageChangeHandler?: any; progressHandler?: any },
 ) {
   if (titleElement.querySelector(".search-trigger")) {
     return;
@@ -21,6 +21,72 @@ export function mountSearchBar(
   const searchButton = document.createElement("div");
   searchButton.className = "search-trigger";
   
+  // Create progress indicator container
+  const progressContainer = document.createElement("div");
+  progressContainer.className = "search-progress-container";
+  progressContainer.style.cssText = "display: flex; align-items: center; gap: 8px; margin-left: 8px; min-width: 120px;";
+  
+  // Create progress bar
+  const progressBarWrapper = document.createElement("div");
+  progressBarWrapper.className = "search-progress-bar-wrapper";
+  progressBarWrapper.style.cssText = "flex: 1; height: 4px; background: rgba(0, 0, 0, 0.1); border-radius: 2px; overflow: hidden; display: none;";
+  
+  const progressBar = document.createElement("div");
+  progressBar.className = "search-progress-bar";
+  progressBar.style.cssText = "height: 100%; background: linear-gradient(90deg, #3b82f6, #2563eb, #3b82f6); transition: width 0.3s ease-out; width: 0%; position: relative;";
+  
+  // Add shimmer effect
+  const shimmer = document.createElement("div");
+  shimmer.style.cssText = "position: absolute; inset: 0; background: linear-gradient(90deg, transparent, rgba(255,255,255,0.3), transparent); animation: shimmer 2s infinite;";
+  progressBar.appendChild(shimmer);
+  progressBarWrapper.appendChild(progressBar);
+  
+  // Create progress text
+  const progressText = document.createElement("span");
+  progressText.className = "search-progress-text";
+  progressText.style.cssText = "font-size: 11px; color: #666; white-space: nowrap; display: none;";
+  
+  progressContainer.appendChild(progressBarWrapper);
+  progressContainer.appendChild(progressText);
+  
+  // Indexing state
+  let isIndexing = false;
+  let completedJobs = 0;
+  let totalJobs = 0;
+  let indexingStatus: string | null = null;
+  
+  const updateProgressDisplay = () => {
+    if (isIndexing && totalJobs > 0) {
+      const percentage = Math.round((completedJobs / totalJobs) * 100);
+      progressBar.style.width = `${Math.max(2, percentage)}%`;
+      progressBarWrapper.style.display = "block";
+      
+      if (indexingStatus) {
+        progressText.textContent = indexingStatus.length > 20 ? indexingStatus.substring(0, 20) + "..." : indexingStatus;
+        progressText.style.display = "block";
+      } else {
+        progressText.textContent = `${completedJobs}/${totalJobs} (${percentage}%)`;
+        progressText.style.display = "block";
+      }
+    } else {
+      progressBarWrapper.style.display = "none";
+      progressText.style.display = "none";
+    }
+  };
+  
+  // Listen for indexing progress events
+  const progressHandler = (event: CustomEvent) => {
+    const { completed, total, indexing, status } = event.detail;
+    completedJobs = completed || 0;
+    totalJobs = total || 0;
+    isIndexing = indexing || false;
+    indexingStatus = status || null;
+    updateProgressDisplay();
+  };
+  
+  window.addEventListener('indexing-progress', progressHandler as EventListener);
+  appRef.progressHandler = progressHandler;
+  
   const updateSearchButtonDisplay = () => {
     searchButton.innerHTML = /* html */ `
       <svg xmlns="http://www.w3.org/2000/svg" width="16" height="16" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round">
@@ -34,6 +100,7 @@ export function mountSearchBar(
 
   updateSearchButtonDisplay();
   titleElement.appendChild(searchButton);
+  titleElement.appendChild(progressContainer);
 
   // Listen for hotkey setting changes
   const handleStorageChange = (changes: any, area: string) => {
@@ -72,7 +139,7 @@ export function mountSearchBar(
   }
 }
 
-export function cleanupSearchBar(appRef: { current: any; storageChangeHandler?: any }) {
+export function cleanupSearchBar(appRef: { current: any; storageChangeHandler?: any; progressHandler?: any }) {
   if (appRef.current) {
     try {
       unmount(appRef.current);
@@ -82,12 +149,24 @@ export function cleanupSearchBar(appRef: { current: any; storageChangeHandler?:
     }
   }
 
+  // Remove progress event listener
+  if (appRef.progressHandler) {
+    window.removeEventListener('indexing-progress', appRef.progressHandler as EventListener);
+    appRef.progressHandler = null;
+  }
+
   // Remove search trigger button
   const searchTrigger = document.querySelector(".search-trigger");
   if (searchTrigger) {
     searchTrigger.remove();
   }
   
+  // Remove progress container
+  const progressContainer = document.querySelector(".search-progress-container");
+  if (progressContainer) {
+    progressContainer.remove();
+  }
+
   // Remove search root
   const searchRoot = document.querySelector("div[data-search-root]");
   if (searchRoot) {

src/plugins/built-in/globalSearch/src/core/styles.css

@@ -69,3 +69,71 @@
 .dark .highlight {
   background-color: rgba(255, 230, 100, 0.4);
 }
+
+@keyframes shimmer {
+  0% {
+    transform: translateX(-100%);
+  }
+  100% {
+    transform: translateX(100%);
+  }
+}
+
+.animate-shimmer {
+  animation: shimmer 2s infinite;
+}
+
+/* Progress indicator next to search trigger */
+.search-progress-container {
+  display: flex;
+  align-items: center;
+  gap: 8px;
+  margin-left: 8px;
+  min-width: 120px;
+  max-width: 200px;
+  height: 32px;
+}
+
+.search-progress-bar-wrapper {
+  flex: 1;
+  height: 4px;
+  background: rgba(0, 0, 0, 0.1);
+  border-radius: 2px;
+  overflow: hidden;
+  display: none;
+  min-width: 60px;
+}
+
+.dark .search-progress-bar-wrapper {
+  background: rgba(255, 255, 255, 0.1);
+}
+
+.search-progress-bar {
+  height: 100%;
+  background: linear-gradient(90deg, #3b82f6, #2563eb, #3b82f6);
+  transition: width 0.3s ease-out;
+  width: 0%;
+  position: relative;
+  border-radius: 2px;
+}
+
+.search-progress-bar::after {
+  content: '';
+  position: absolute;
+  inset: 0;
+  background: linear-gradient(90deg, transparent, rgba(255,255,255,0.3), transparent);
+  animation: shimmer 2s infinite;
+  border-radius: 2px;
+}
+
+.search-progress-text {
+  font-size: 11px;
+  color: #666;
+  white-space: nowrap;
+  display: none;
+  font-weight: 500;
+}
+
+.dark .search-progress-text {
+  color: #999;
+}

src/plugins/built-in/globalSearch/src/indexing/actions.ts

@@ -59,17 +59,132 @@ export const actionMap: Record<string, ActionHandler<any>> = {
   }) as ActionHandler<any>,
 
   assessment: (async (item: IndexItem & { metadata: AssessmentMetadata }) => {
-    if (item.metadata.isMessageBased) {
+    // Deep clone the entire item to avoid Firefox XrayWrapper issues
+    // Firefox XrayWrapper prevents direct access to nested properties
+    let itemClone: IndexItem & { metadata: AssessmentMetadata };
+    let metadata: AssessmentMetadata;
+    
+    try {
+      // First try to clone the entire item
+      itemClone = JSON.parse(JSON.stringify(item));
+      metadata = itemClone.metadata || {};
+    } catch (e) {
+      console.warn("[Assessment Action] Failed to clone item, trying to clone metadata separately:", e);
+      try {
+        // If full clone fails, try cloning just metadata
+        metadata = JSON.parse(JSON.stringify(item.metadata || {}));
+        itemClone = { ...item, metadata };
+      } catch (e2) {
+        console.warn("[Assessment Action] Failed to clone metadata, using direct access:", e2);
+        itemClone = item;
+        metadata = item.metadata || {} as AssessmentMetadata;
+      }
+    }
+    
+    // Try to extract metadata values using multiple methods to handle XrayWrapper
+    const getMetadataValue = (key: string, altKey?: string): any => {
+      try {
+        // Try direct access first
+        const value = metadata[key];
+        if (value !== undefined && value !== null) {
+          return value;
+        }
+        if (altKey) {
+          const altValue = metadata[altKey];
+          if (altValue !== undefined && altValue !== null) {
+            return altValue;
+          }
+        }
+        // Try accessing via Object.keys iteration (works around XrayWrapper)
+        try {
+          const keys = Object.keys(metadata);
+          for (const k of keys) {
+            if (k === key || k === altKey) {
+              const val = metadata[k];
+              if (val !== undefined && val !== null) {
+                return val;
+              }
+            }
+          }
+        } catch (e) {
+          // Object.keys might fail on XrayWrapper, that's okay
+        }
+        return undefined;
+      } catch (e) {
+        console.warn(`[Assessment Action] Failed to access metadata.${key}:`, e);
+        return undefined;
+      }
+    };
+    
+    if (getMetadataValue('isMessageBased')) {
       window.location.hash = `#?page=/messages`;
 
       await waitForElm('[class*="Viewer__Viewer___"] > div', true, 20);
 
       // Select the specific direct message
       ReactFiber.find('[class*="Viewer__Viewer___"] > div').setState({
-        selected: new Set([item.metadata.messageId]),
+        selected: new Set([getMetadataValue('messageId')]),
       });
     } else {
-      window.location.hash = `#?page=/assessments&id=${item.metadata.assessmentId}`;
+      // Extract values - check both camelCase and PascalCase, and try multiple access methods
+      let programmeId = getMetadataValue('programmeId', 'programmeID');
+      let metaclassId = getMetadataValue('metaclassId', 'metaclassID');
+      let assessmentId = getMetadataValue('assessmentId', 'assessmentID');
+      
+      // Fallback: try to extract assessmentId from item ID if metadata is missing
+      if ((assessmentId === undefined || assessmentId === null) && itemClone.id && itemClone.id.startsWith('assignment-')) {
+        const extractedId = itemClone.id.replace('assignment-', '');
+        assessmentId = Number(extractedId) || extractedId;
+        console.log("[Assessment Action] Extracted assessmentId from item ID:", assessmentId);
+      }
+      
+      // Convert to numbers, but preserve 0 as valid
+      if (programmeId !== undefined && programmeId !== null && programmeId !== '') {
+        const num = Number(programmeId);
+        programmeId = isNaN(num) ? programmeId : num;
+      }
+      if (metaclassId !== undefined && metaclassId !== null && metaclassId !== '') {
+        const num = Number(metaclassId);
+        metaclassId = isNaN(num) ? metaclassId : num;
+      }
+      if (assessmentId !== undefined && assessmentId !== null && assessmentId !== '') {
+        const num = Number(assessmentId);
+        assessmentId = isNaN(num) ? assessmentId : num;
+      }
+      
+      // Check if values exist (including 0, which is a valid ID)
+      // Use typeof check to properly handle 0
+      const hasProgrammeId = programmeId !== undefined && programmeId !== null && programmeId !== '' && typeof programmeId === 'number';
+      const hasMetaclassId = metaclassId !== undefined && metaclassId !== null && metaclassId !== '' && typeof metaclassId === 'number';
+      const hasAssessmentId = assessmentId !== undefined && assessmentId !== null && assessmentId !== '' && typeof assessmentId === 'number';
+      
+     
+      
+      if (hasProgrammeId && hasMetaclassId && hasAssessmentId) {
+        const url = `#?page=/assessments/${programmeId}:${metaclassId}&item=${assessmentId}`;
+        console.log("[Assessment Action] ✅ Navigating to:", url);
+        window.location.hash = url;
+      } else {
+        // Fallback: try to navigate to assessments page if metadata is incomplete
+        console.error("[Assessment Action] ❌ Missing required metadata:", {
+          programmeId,
+          metaclassId,
+          assessmentId,
+          hasProgrammeId,
+          hasMetaclassId,
+          hasAssessmentId,
+          metadataKeys: Object.keys(metadata),
+          metadataString: JSON.stringify(metadata),
+          itemId: itemClone.id,
+        });
+        // If we at least have an assessmentId, try to navigate to the general assessments page
+        if (hasAssessmentId) {
+          window.location.hash = `#?page=/assessments/upcoming&item=${assessmentId}`;
+        } else {
+          console.warn("[Assessment Action] No valid assessment ID, redirecting to upcoming");
+          window.location.hash = `#?page=/assessments/upcoming`;
+        }
+      }
     }
   }) as ActionHandler<any>,
 

src/plugins/built-in/globalSearch/src/indexing/db.ts

@@ -213,25 +213,54 @@ export async function clear(store: string): Promise<void> {
 }
 
 export async function resetDatabase(): Promise<void> {
+  // Close cached database connection
   if (cachedDb) {
-    cachedDb.close();
+    try {
+      cachedDb.close();
+    } catch (e) {
+      console.warn("[DB] Error closing cached database:", e);
+    }
     cachedDb = null;
   }
 
+  // Close pending database promise
   if (dbPromise) {
     try {
       const db = await dbPromise;
       db.close();
-    } catch (e) {}
+    } catch (e) {
+      // Database might not be open yet, that's okay
+    }
     dbPromise = null;
   }
 
+  // Wait a bit for connections to fully close
+  await new Promise(resolve => setTimeout(resolve, 100));
+
   return new Promise((resolve, reject) => {
     const req = indexedDB.deleteDatabase(DB_NAME);
     req.onsuccess = () => {
       localStorage.removeItem(VERSION_KEY);
       resolve();
     };
-    req.onerror = () => reject(req.error);
+    req.onerror = () => {
+      console.error("[DB] Error deleting database:", req.error);
+      reject(req.error);
+    };
+    req.onblocked = () => {
+      console.warn("[DB] Database deletion blocked - waiting for connections to close");
+      // Wait a bit longer and try again
+      setTimeout(() => {
+        const retryReq = indexedDB.deleteDatabase(DB_NAME);
+        retryReq.onsuccess = () => {
+          localStorage.removeItem(VERSION_KEY);
+          resolve();
+        };
+        retryReq.onerror = () => reject(retryReq.error);
+        retryReq.onblocked = () => {
+          reject(new Error(`Database is still open. Please close other tabs/windows and try again.`));
+        };
+      }, 500);
+    };
   });
 }

src/plugins/built-in/globalSearch/src/indexing/indexer.ts

@@ -396,18 +396,34 @@ export async function runIndexing(): Promise<void> {
   stopHeartbeat();
 
   allItemsInPrimaryStores = await loadAllStoredItems();
-  allItemsInPrimaryStores.forEach(item => {
+  // Create new objects to avoid XrayWrapper issues in Firefox
-    const jobDef = jobs[item.category] || Object.values(jobs).find(j => j.id === item.category) || jobs[item.renderComponentId];
+  const itemsWithComponents = allItemsInPrimaryStores.map(item => {
-    if (jobDef) {
+    try {
-        const renderComponent = renderComponentMap[jobDef.renderComponentId];
+      const jobDef = jobs[item.category] || Object.values(jobs).find(j => j.id === item.category) || jobs[item.renderComponentId];
-        if (renderComponent) {
+      let renderComponent = item.renderComponent;
-          item.renderComponent = renderComponent;
+      if (jobDef) {
-        }
+        renderComponent = renderComponentMap[jobDef.renderComponentId] || renderComponent;
-    } else if (renderComponentMap[item.renderComponentId]) {
+      } else if (renderComponentMap[item.renderComponentId]) {
-        item.renderComponent = renderComponentMap[item.renderComponentId];
+        renderComponent = renderComponentMap[item.renderComponentId];
+      }
+      // Deep clone to avoid Firefox XrayWrapper issues with nested objects like metadata
+      // Use JSON serialization to ensure all nested properties are accessible
+      try {
+        const cloned = JSON.parse(JSON.stringify(item));
+        cloned.renderComponent = renderComponent;
+        return cloned;
+      } catch (e) {
+        // Fallback to shallow copy if deep clone fails
+        console.warn("[Indexer] Failed to deep clone item, using shallow copy:", e);
+        return { ...item, renderComponent };
+      }
+    } catch (error) {
+      // Fallback: return item as-is if modification fails (Firefox XrayWrapper)
+      console.warn("[Indexer] Failed to add render component to item (Firefox XrayWrapper):", error);
+      return item;
     }
   });
-  loadDynamicItems(allItemsInPrimaryStores);
+  loadDynamicItems(itemsWithComponents);
   window.dispatchEvent(new Event("dynamic-items-updated"));
 }
 

src/plugins/built-in/globalSearch/src/indexing/jobs.ts

@@ -3,10 +3,12 @@ import { messagesJob } from "./jobs/messages";
 import { notificationsJob } from "./jobs/notifications";
 import { forumsJob } from "./jobs/forums";
 import { subjectsJob } from "./jobs/subjects";
+import { assignmentsJob } from "./jobs/assignments";
 
 export const jobs: Record<string, Job> = {
   messages: messagesJob,
   notifications: notificationsJob,
   forums: forumsJob,
   subjects: subjectsJob,
+  assignments: assignmentsJob,
 };

src/plugins/built-in/globalSearch/src/indexing/jobs/assignments.ts

@@ -0,0 +1,369 @@
+import type { IndexItem, Job } from "../types";
+
+const fetchJSON = async (url: string, body: any) => {
+  const res = await fetch(`${location.origin}${url}`, {
+    method: "POST",
+    credentials: "include",
+    headers: { "Content-Type": "application/json; charset=utf-8" },
+    body: JSON.stringify(body),
+  });
+  return res.json();
+};
+
+const fetchUpcomingAssessments = async (student: number = 69) => {
+  try {
+    const res = await fetchJSON("/seqta/student/assessment/list/upcoming?", {
+      student,
+    });
+    // Match analytics.rs: payload is an array, return empty array if not found
+    return Array.isArray(res.payload) ? res.payload : [];
+  } catch (e) {
+    console.error("[Assignments job] Failed to fetch upcoming assessments:", e);
+    return [];
+  }
+};
+
+const fetchSubjects = async () => {
+  try {
+    const res = await fetchJSON("/seqta/student/load/subjects?", {});
+    return res.payload
+      ?.filter((s: any) => s.active === 1)
+      ?.flatMap((s: any) => s.subjects) || [];
+  } catch (e) {
+    console.error("[Assignments job] Failed to fetch subjects:", e);
+    return [];
+  }
+};
+
+const fetchPastAssessments = async (student: number = 69, subjects: any[]) => {
+  const map: Record<number, any> = {};
+  
+  // Fetch past assessments for all subjects in parallel (like assessmentsOverview does)
+  // This is much faster than sequential fetching
+  await Promise.all(
+    subjects.map(async (subject) => {
+      try {
+        // Match analytics.rs exactly: parameter order is programme, metaclass, student
+        const res = await fetchJSON("/seqta/student/assessment/list/past?", {
+          programme: subject.programme,
+          metaclass: subject.metaclass,
+          student,
+        });
+        
+        // Past assessments API can return data in payload.tasks OR payload.pending (or both)
+        // Based on analytics.rs fetch_past_assessments, we need to check both arrays
+        const processAssessment = (assessment: any) => {
+          if (assessment && assessment.id) {
+            // Ensure programme and metaclass are included from the subject
+            // Use the assessment's IDs if available, otherwise fall back to subject's
+            map[assessment.id] = {
+              ...assessment,
+              programme: assessment.programme || assessment.programmeID || subject.programme,
+              programmeID: assessment.programmeID || assessment.programme || subject.programme,
+              metaclass: assessment.metaclass || assessment.metaclassID || subject.metaclass,
+              metaclassID: assessment.metaclassID || assessment.metaclass || subject.metaclass,
+            };
+          }
+        };
+        
+        // Match analytics.rs: Check both pending and tasks arrays
+        // Check for pending array first (matching Rust code order)
+        if (res.payload?.pending && Array.isArray(res.payload.pending)) {
+          res.payload.pending.forEach(processAssessment);
+        }
+        
+        // Check for tasks array
+        if (res.payload?.tasks && Array.isArray(res.payload.tasks)) {
+          res.payload.tasks.forEach(processAssessment);
+        }
+      } catch (e) {
+        console.warn(`[Assignments job] Failed to fetch past assessments for subject ${subject.code || subject.subject || 'unknown'}:`, e);
+      }
+    })
+  );
+  
+  return Object.values(map);
+};
+
+export const assignmentsJob: Job = {
+  id: "assignments",
+  label: "Assignments",
+  renderComponentId: "assessment",
+  frequency: { type: "expiry", afterMs: 1000 * 60 * 60 * 24 }, // Daily
+  
+  boostCriteria: (item, searchTerm) => {
+    if (searchTerm === "") {
+      return -100;
+    }
+
+    let score = 0;
+    
+    // Boost upcoming assignments
+    if (item.metadata.dueDate) {
+      const dueDate = new Date(item.metadata.dueDate).getTime();
+      const now = Date.now();
+      const daysUntilDue = (dueDate - now) / (1000 * 60 * 60 * 24);
+      
+      if (daysUntilDue >= 0 && daysUntilDue <= 7) {
+        score += 0.05; // Boost assignments due within a week
+      }
+      if (daysUntilDue < 0) {
+        score -= 0.1; // Penalty for overdue assignments
+      }
+    }
+    
+    // Boost if submitted
+    if (item.metadata.submitted) {
+      score += 0.02;
+    }
+    
+    return score;
+  },
+
+  run: async (ctx) => {
+    // Don't filter by existing IDs - we want to process ALL assessments (both new and old)
+    // to ensure metadata is up-to-date and all past assignments are indexed
+    const existingItems = await ctx.getStoredItems("assignments");
+    const existingIds = new Set(existingItems.map((i) => i.id));
+
+    const student = 69; // TODO: Get from context if available
+    
+    console.debug("[Assignments job] Starting indexing - fetching all assessments (upcoming and past)...");
+    
+    // Fetch data in parallel
+    const [upcoming, subjects] = await Promise.all([
+      fetchUpcomingAssessments(student),
+      fetchSubjects(),
+    ]);
+
+    console.debug(`[Assignments job] Fetched ${upcoming.length} upcoming assessments and ${subjects.length} subjects`);
+
+    // Fetch past assessments for ALL subjects to ensure we get all historical assignments
+    const past = await fetchPastAssessments(student, subjects);
+    
+    console.debug(`[Assignments job] Fetched ${past.length} past assessments`);
+    
+    // Create a lookup map from subject code to programme/metaclass
+    const subjectLookup = new Map<string, { programme: number; metaclass: number }>();
+    subjects.forEach((s: any) => {
+      if (s.code && s.programme && s.metaclass) {
+        subjectLookup.set(s.code, { programme: s.programme, metaclass: s.metaclass });
+      }
+    });
+    
+    // Combine and deduplicate
+    const allAssessments = new Map<number, any>();
+    
+    upcoming.forEach((a: any) => {
+      if (a && a.id) {
+        // Prioritize capital ID fields (programmeID, metaclassID) as that's what the API returns
+        let programme = a.programmeID || a.programme;
+        let metaclass = a.metaclassID || a.metaclass;
+        
+        // If missing, try to get from subject lookup
+        if ((!programme || !metaclass) && a.code) {
+          const subjectInfo = subjectLookup.get(a.code);
+          if (subjectInfo) {
+            programme = programme || subjectInfo.programme;
+            metaclass = metaclass || subjectInfo.metaclass;
+          }
+        }
+        
+        allAssessments.set(a.id, {
+          ...a,
+          programme,
+          metaclass,
+          programmeID: programme, // Ensure both formats are available
+          metaclassID: metaclass,
+          isUpcoming: true,
+        });
+      }
+    });
+    
+    past.forEach((a: any) => {
+      if (a && a.id) {
+        // Prioritize capital ID fields (programmeID, metaclassID) as that's what the API returns
+        let programme = a.programmeID || a.programme;
+        let metaclass = a.metaclassID || a.metaclass;
+        
+        const existing = allAssessments.get(a.id);
+        if (existing) {
+          // Merge past assessment data, ensuring programme/metaclass are preserved
+          // Use existing values if new ones are missing
+          programme = programme || existing.programme || existing.programmeID;
+          metaclass = metaclass || existing.metaclass || existing.metaclassID;
+          
+          Object.assign(existing, {
+            ...a,
+            programme,
+            metaclass,
+            programmeID: programme,
+            metaclassID: metaclass,
+          });
+        } else {
+          allAssessments.set(a.id, { 
+            ...a, 
+            programme,
+            metaclass,
+            programmeID: programme,
+            metaclassID: metaclass,
+            isUpcoming: false 
+          });
+        }
+      }
+    });
+
+    const items: IndexItem[] = [];
+    const processedIds = new Set<string>();
+
+    // Process assessments in batches to avoid overwhelming the API
+    const assessmentArray = Array.from(allAssessments.values());
+    const pastCount = assessmentArray.filter(a => !a.isUpcoming).length;
+    const upcomingCount = assessmentArray.filter(a => a.isUpcoming).length;
+    console.debug(`[Assignments job] Processing ${assessmentArray.length} total assessments (${upcomingCount} upcoming, ${pastCount} past)`);
+    const batchSize = 15; // Increased batch size for better performance
+    
+    // Skip fetching assessment details - the API endpoint doesn't exist or returns 404
+    // Details are optional and not critical for search functionality
+    
+    // Process ALL assessments (both upcoming and past) to ensure everything is indexed
+    for (let i = 0; i < assessmentArray.length; i += batchSize) {
+      const batch = assessmentArray.slice(i, i + batchSize);
+      
+      const batchItems = await Promise.all(
+        batch.map(async (assessment) => {
+          const id = `assignment-${assessment.id}`;
+          
+          // Skip if already processed in this batch
+          if (processedIds.has(id)) {
+            return null;
+          }
+          
+          processedIds.add(id);
+          
+          // Process ALL assessments (both new and existing, upcoming and past)
+          // This ensures all historical assignments are indexed and metadata is up-to-date
+
+          // Skip fetching details - API endpoint doesn't exist
+          const description = "";
+
+          const subjectName = assessment.subject || assessment.code || "Unknown Subject";
+          const dueDate = assessment.due ? new Date(assessment.due).getTime() : null;
+          
+          // Prioritize capital ID fields (programmeID, metaclassID) as that's what the API returns
+          const programmeId = assessment.programmeID || assessment.programme;
+          const metaclassId = assessment.metaclassID || assessment.metaclass;
+          
+          // Validate that we have the required IDs for navigation
+          if (!programmeId || !metaclassId || !assessment.id) {
+            console.warn(`[Assignments job] Skipping assignment ${assessment.id} - missing required IDs:`, {
+              programmeId,
+              metaclassId,
+              assessmentId: assessment.id,
+              programmeID: assessment.programmeID,
+              metaclassID: assessment.metaclassID,
+              programme: assessment.programme,
+              metaclass: assessment.metaclass,
+              assessment,
+            });
+            return null;
+          }
+          
+          // Convert to numbers, preserving 0 as valid
+          let finalProgrammeId: number | undefined;
+          let finalMetaclassId: number | undefined;
+          
+          if (programmeId !== undefined && programmeId !== null && programmeId !== '') {
+            const num = Number(programmeId);
+            finalProgrammeId = isNaN(num) ? undefined : num;
+          }
+          
+          if (metaclassId !== undefined && metaclassId !== null && metaclassId !== '') {
+            const num = Number(metaclassId);
+            finalMetaclassId = isNaN(num) ? undefined : num;
+          }
+          
+          // Final validation - check for actual numbers (including 0)
+          if (finalProgrammeId === undefined || finalMetaclassId === undefined || !assessment.id) {
+            console.error(`[Assignments job] ❌ Skipping assignment ${assessment.id} - invalid IDs after conversion:`, {
+              programmeId: finalProgrammeId,
+              metaclassId: finalMetaclassId,
+              assessmentId: assessment.id,
+              rawProgrammeId: programmeId,
+              rawMetaclassId: metaclassId,
+              assessment,
+            });
+            return null;
+          }
+          
+          const item: IndexItem = {
+            id,
+            text: assessment.title || assessment.name || "Untitled Assignment",
+            category: "assignments",
+            content: `${description}\nSubject: ${subjectName}\nDue: ${assessment.due || "No due date"}`.trim(),
+            dateAdded: dueDate || Date.now(),
+            metadata: {
+              assessmentId: assessment.id,
+              assessmentID: assessment.id, // Store both variants for compatibility
+              subject: subjectName,
+              subjectCode: assessment.code,
+              dueDate: assessment.due,
+              programmeId: finalProgrammeId,
+              programmeID: finalProgrammeId, // Store both variants for compatibility
+              metaclassId: finalMetaclassId,
+              metaclassID: finalMetaclassId, // Store both variants for compatibility
+              submitted: assessment.submitted || false,
+              isUpcoming: assessment.isUpcoming || false,
+              term: assessment.term,
+              timestamp: assessment.due || new Date().toISOString(), // Required by AssessmentMetadata interface
+            },
+            actionId: "assessment",
+            renderComponentId: "assessment",
+          };
+          
+          console.debug(`[Assignments job] ✅ Created item for assignment ${assessment.id}:`, {
+            id: item.id,
+            programmeId: item.metadata.programmeId,
+            programmeID: item.metadata.programmeID,
+            metaclassId: item.metadata.metaclassId,
+            metaclassID: item.metadata.metaclassID,
+            assessmentId: item.metadata.assessmentId,
+            assessmentID: item.metadata.assessmentID,
+          });
+
+          return item;
+        })
+      );
+      
+      // Filter out nulls and add to items
+      batchItems.forEach(item => {
+        if (item) {
+          items.push(item);
+        }
+      });
+      
+      // Small delay between batches to avoid rate limiting
+      if (i + batchSize < assessmentArray.length) {
+        await new Promise(resolve => setTimeout(resolve, 50)); // Reduced delay
+      }
+    }
+
+    const newItemsCount = items.filter(item => !existingIds.has(item.id)).length;
+    const updatedItemsCount = items.length - newItemsCount;
+    console.debug(`[Assignments job] Indexed ${items.length} assignment items (${newItemsCount} new, ${updatedItemsCount} updated)`);
+    return items;
+  },
+
+  purge: (items) => {
+    // Keep ALL assignments - don't purge old ones as users may want to search for them
+    // Only remove items that are truly invalid (missing required metadata)
+    return items.filter((i) => {
+      // Keep all items that have valid metadata
+      return i.metadata && 
+             i.metadata.assessmentId && 
+             i.metadata.programmeId !== undefined && 
+             i.metadata.metaclassId !== undefined;
+    });
+  },
+};
+

src/plugins/built-in/globalSearch/src/indexing/jobs/messages.ts

@@ -604,22 +604,34 @@ export const messagesJob: Job = {
       if (processedItems.length > 0) {
         try {
           const currentItems = await loadAllStoredItems();
-          currentItems.forEach((item) => {
+          // Create new objects to avoid XrayWrapper issues in Firefox
-            const jobDef =
+          const itemsWithComponents = currentItems.map((item) => {
-              jobs[item.category] ||
+            try {
-              Object.values(jobs).find((j) => j.id === item.category) ||
+              const jobDef =
-              jobs[item.renderComponentId];
+                jobs[item.category] ||
-            if (jobDef) {
+                Object.values(jobs).find((j) => j.id === item.category) ||
-              const renderComponent =
+                jobs[item.renderComponentId];
-                renderComponentMap[jobDef.renderComponentId];
+              let renderComponent = item.renderComponent;
-              if (renderComponent) {
+              if (jobDef) {
-                item.renderComponent = renderComponent;
+                renderComponent = renderComponentMap[jobDef.renderComponentId] || renderComponent;
+              } else if (renderComponentMap[item.renderComponentId]) {
+                renderComponent = renderComponentMap[item.renderComponentId];
               }
-            } else if (renderComponentMap[item.renderComponentId]) {
+              // Deep clone to avoid Firefox XrayWrapper issues with nested objects like metadata
-              item.renderComponent = renderComponentMap[item.renderComponentId];
+              try {
+                const cloned = JSON.parse(JSON.stringify(item));
+                cloned.renderComponent = renderComponent;
+                return cloned;
+              } catch (e) {
+                // Fallback to shallow copy if deep clone fails
+                return { ...item, renderComponent };
+              }
+            } catch (error) {
+              // Fallback: return item as-is if modification fails (Firefox XrayWrapper)
+              return item;
             }
           });
-          loadDynamicItems(currentItems);
+          loadDynamicItems(itemsWithComponents);
           window.dispatchEvent(
             new CustomEvent("dynamic-items-updated", {
               detail: {

src/plugins/built-in/globalSearch/src/indexing/jobs/notifications.ts

@@ -372,23 +372,34 @@ export const notificationsJob: Job = {
         if (items.length > 0) {
           try {
             const currentItems = await loadAllStoredItems();
-            currentItems.forEach((item) => {
+            // Create new objects to avoid XrayWrapper issues in Firefox
-              const jobDef =
+            const itemsWithComponents = currentItems.map((item) => {
-                jobs[item.category] ||
+              try {
-                Object.values(jobs).find((j) => j.id === item.category) ||
+                const jobDef =
-                jobs[item.renderComponentId];
+                  jobs[item.category] ||
-              if (jobDef) {
+                  Object.values(jobs).find((j) => j.id === item.category) ||
-                const renderComponent =
+                  jobs[item.renderComponentId];
-                  renderComponentMap[jobDef.renderComponentId];
+                let renderComponent = item.renderComponent;
-                if (renderComponent) {
+                if (jobDef) {
-                  item.renderComponent = renderComponent;
+                  renderComponent = renderComponentMap[jobDef.renderComponentId] || renderComponent;
+                } else if (renderComponentMap[item.renderComponentId]) {
+                  renderComponent = renderComponentMap[item.renderComponentId];
                 }
-              } else if (renderComponentMap[item.renderComponentId]) {
+                // Deep clone to avoid Firefox XrayWrapper issues with nested objects like metadata
-                item.renderComponent =
+                try {
-                  renderComponentMap[item.renderComponentId];
+                  const cloned = JSON.parse(JSON.stringify(item));
+                  cloned.renderComponent = renderComponent;
+                  return cloned;
+                } catch (e) {
+                  // Fallback to shallow copy if deep clone fails
+                  return { ...item, renderComponent };
+                }
+              } catch (error) {
+                // Fallback: return item as-is if modification fails (Firefox XrayWrapper)
+                return item;
               }
             });
-            loadDynamicItems(currentItems);
+            loadDynamicItems(itemsWithComponents);
             window.dispatchEvent(
               new CustomEvent("dynamic-items-updated", {
                 detail: {

src/plugins/built-in/globalSearch/src/indexing/worker/vectorWorker.ts

@@ -3,9 +3,24 @@ import type { IndexItem } from "../types";
 
 let vectorIndex: EmbeddingIndex | null = null;
 let isInitialized = false;
+let initializationFailed = false;
 let currentAbortController: AbortController | null = null;
 let loadedItemIds = new Set<string>();
 
+// Detect Firefox in worker context
+function isFirefoxWorker(): boolean {
+  try {
+    // Check for Firefox-specific APIs or user agent
+    if (typeof navigator !== "undefined") {
+      return navigator.userAgent.toLowerCase().includes("firefox");
+    }
+    // In worker context, check for Firefox-specific behavior
+    return false;
+  } catch {
+    return false;
+  }
+}
+
 let streamingSession: {
   isActive: boolean;
   totalExpected: number;
@@ -21,6 +36,16 @@ async function initWorker() {
     console.debug("Vector worker already initialized.");
     return;
   }
+  
+  // Skip initialization in Firefox
+  if (isFirefoxWorker()) {
+    console.debug("[Vector Worker] Vector search not supported in Firefox - skipping initialization");
+    isInitialized = true;
+    initializationFailed = true;
+    vectorIndex = null;
+    return;
+  }
+  
   console.debug("Initializing vector worker...");
   try {
     await initializeModel();
@@ -48,8 +73,9 @@ async function initWorker() {
     isInitialized = true;
     console.debug("Vector worker initialized successfully.");
   } catch (e) {
-    console.error("Failed to initialize vector worker:", e);
+    console.warn("[Vector Worker] Failed to initialize vector worker (will use text search only):", e);
     isInitialized = true;
+    initializationFailed = true;
     vectorIndex = null;
   }
 }
@@ -80,18 +106,29 @@ async function startStreamingSession(
   totalExpected: number,
   batchSize: number = 5,
 ) {
+  if (initializationFailed || isFirefoxWorker()) {
+    self.postMessage({
+      type: "progress",
+      data: {
+        status: "complete",
+        message: "Vector search not available in Firefox - using text search only",
+      },
+    });
+    return;
+  }
+  
   if (!vectorIndex) {
     console.warn(
       "Streaming requested but vector index not ready. Attempting init.",
     );
     await initWorker();
-    if (!vectorIndex) {
+    if (!vectorIndex || initializationFailed) {
       self.postMessage({
         type: "progress",
         data: {
-          status: "error",
+          status: "complete",
           message:
-            "Vector index not available for streaming after init attempt.",
+            "Vector index not available - using text search only",
         },
       });
       return;
@@ -306,18 +343,29 @@ async function endStreamingSession() {
 async function processItems(items: IndexItem[], signal: AbortSignal) {
   console.debug("Worker received process request.");
 
+  if (initializationFailed || isFirefoxWorker()) {
+    self.postMessage({
+      type: "progress",
+      data: {
+        status: "complete",
+        message: "Vector search not available - using text search only",
+      },
+    });
+    return;
+  }
+
   if (!vectorIndex) {
     console.warn(
       "Processing requested but vector index not ready. Attempting init.",
     );
     await initWorker();
-    if (!vectorIndex) {
+    if (!vectorIndex || initializationFailed) {
       self.postMessage({
         type: "progress",
         data: {
-          status: "error",
+          status: "complete",
           message:
-            "Vector index not available for processing after init attempt.",
+            "Vector index not available - using text search only",
         },
       });
       return;

src/plugins/built-in/globalSearch/src/indexing/worker/vectorWorkerManager.ts

@@ -1,5 +1,6 @@
 import { refreshVectorCache } from "../../search/vector/vectorSearch";
 import type { IndexItem } from "../types";
+import { isVectorSearchSupported } from "../../utils/browserDetection";
 import vectorWorker from "./vectorWorker.ts?inlineWorker";
 
 export type ProgressCallback = (data: {
@@ -42,6 +43,13 @@ export class VectorWorkerManager {
   }
 
   private async initWorker(): Promise<void> {
+    // Skip initialization if vector search is not supported (e.g., Firefox)
+    if (!isVectorSearchSupported()) {
+      console.debug("[VectorWorkerManager] Vector search not supported - skipping worker initialization");
+      this.isInitialized = false;
+      return Promise.resolve();
+    }
+
     if (this.isInitialized) return Promise.resolve();
     if (this.readyPromise) return this.readyPromise;
 
@@ -234,6 +242,17 @@ export class VectorWorkerManager {
   }
 
   async processItems(items: IndexItem[], onProgress?: ProgressCallback) {
+    // Skip if vector search is not supported
+    if (!isVectorSearchSupported()) {
+      if (onProgress) {
+        onProgress({
+          status: "complete",
+          message: "Vector search not available - using text search only"
+        });
+      }
+      return;
+    }
+
     // Only initialize worker if we actually have items to process
     if (items.length === 0) {
       if (onProgress) {
@@ -298,6 +317,18 @@ export class VectorWorkerManager {
     batchSize: number = 10,
     jobId?: string,
   ): Promise<void> {
+    // Skip if vector search is not supported
+    if (!isVectorSearchSupported()) {
+      console.debug("[VectorWorker] Vector search not supported - skipping streaming session");
+      if (onProgress) {
+        onProgress({
+          status: "complete",
+          message: "Vector search not available - using text search only",
+        });
+      }
+      return;
+    }
+
     // Only initialize if we expect items to process
     if (totalExpectedItems === 0) {
       console.debug("[VectorWorker] No items expected, not starting streaming session");

src/plugins/built-in/globalSearch/src/search/hybridSearch.ts

@@ -0,0 +1,280 @@
+import type { IndexItem } from "../indexing/types";
+import type { CombinedResult } from "../core/types";
+import { searchVectors, type VectorSearchResult } from "./vector/vectorSearch";
+import { jobs } from "../indexing/jobs";
+
+/**
+ * Hybrid Search Implementation
+ * 
+ * Flow:
+ * 1. BM25 (Fuse.js) gets top N results fast
+ * 2. Vector search reranks by semantic similarity
+ * 3. Apply optional boosting (recency, popularity, tags)
+ */
+
+export interface HybridSearchOptions {
+  /** Maximum number of BM25 results to retrieve before reranking */
+  bm25TopK?: number;
+  /** Maximum number of final results to return */
+  finalLimit?: number;
+  /** Whether to apply recency boost */
+  recencyBoost?: boolean;
+  /** Weight for BM25 scores (0-1) */
+  bm25Weight?: number;
+  /** Weight for vector similarity scores (0-1) */
+  vectorWeight?: number;
+  /** Weight for recency boost */
+  recencyWeight?: number;
+}
+
+const DEFAULT_OPTIONS: Required<HybridSearchOptions> = {
+  bm25TopK: 50, // Get top 50 from BM25, then rerank
+  finalLimit: 10,
+  recencyBoost: true,
+  bm25Weight: 0.4, // 40% BM25, 60% vector
+  vectorWeight: 0.6,
+  recencyWeight: 0.1,
+};
+
+/**
+ * Normalizes a score to 0-1 range
+ */
+function normalizeScore(score: number, min: number, max: number): number {
+  if (max === min) return 0.5;
+  return Math.max(0, Math.min(1, (score - min) / (max - min)));
+}
+
+/**
+ * Calculates recency boost based on item age
+ */
+function calculateRecencyBoost(item: IndexItem, now: number): number {
+  const ageInDays = (now - item.dateAdded) / (1000 * 60 * 60 * 24);
+  // Exponential decay: newer items get higher boost
+  // Items from today get boost of 1, items from 30 days ago get ~0.03
+  return 1 / (1 + ageInDays / 7); // Half-life of 7 days
+}
+
+/**
+ * Calculates popularity boost (can be extended with click tracking, etc.)
+ */
+function calculatePopularityBoost(item: IndexItem): number {
+  // For now, boost based on category and metadata
+  let boost = 0;
+  
+  // Boost assignments/assessments
+  if (item.category === "assignments") {
+    boost += 0.1;
+  }
+  
+  // Boost upcoming items
+  if (item.metadata?.isUpcoming) {
+    boost += 0.15;
+  }
+  
+  // Boost items with subject codes (more structured)
+  if (item.metadata?.subjectCode) {
+    boost += 0.05;
+  }
+  
+  return Math.min(boost, 0.3); // Cap at 0.3
+}
+
+/**
+ * Reranks BM25 results using vector search
+ */
+export async function hybridSearch(
+  bm25Results: CombinedResult[],
+  query: string,
+  options: HybridSearchOptions = {},
+): Promise<CombinedResult[]> {
+  const opts = { ...DEFAULT_OPTIONS, ...options };
+  const trimmedQuery = query.trim().toLowerCase();
+  
+  // If no BM25 results, return empty
+  if (bm25Results.length === 0) {
+    return [];
+  }
+  
+  // Limit BM25 results to top K
+  const topBm25Results = bm25Results.slice(0, opts.bm25TopK);
+  
+  // Get vector search results for reranking
+  // We'll search the full index and then filter to our BM25 results
+  let vectorResults: VectorSearchResult[] = [];
+  
+  if (trimmedQuery.length > 2) {
+    try {
+      // Get more vector results than BM25 results to ensure coverage
+      // This allows us to find semantic matches that BM25 might have missed
+      const vectorSearchResults = await searchVectors(trimmedQuery, opts.bm25TopK * 2);
+      
+      // Create a map of item ID to vector similarity
+      const vectorMap = new Map<string, number>();
+      vectorSearchResults.forEach(v => {
+        // Use the highest similarity if item appears multiple times
+        const existing = vectorMap.get(v.object.id);
+        if (!existing || v.similarity > existing) {
+          vectorMap.set(v.object.id, v.similarity);
+        }
+      });
+      
+      // Now rerank BM25 results with vector scores
+      const now = Date.now();
+      
+      const rerankedResults = topBm25Results.map(result => {
+        const item = result.item;
+        
+        // Normalize BM25 score to 0-1
+        // Fuse.js scores: lower is better (0 = perfect match)
+        // We need to invert: higher score = better match
+        // Result.score is typically 0-100, where higher = better
+        // So we normalize it to 0-1
+        const normalizedBm25Score = Math.max(0, Math.min(1, result.score / 100));
+        
+        // Get vector similarity (0-1, already normalized)
+        // If item wasn't in vector results, use a default low score
+        const vectorSimilarity = vectorMap.get(item.id) || 0.3; // Default to 0.3 if not found
+        
+        // Calculate recency boost (0-1 range)
+        const recencyBoost = opts.recencyBoost 
+          ? calculateRecencyBoost(item, now) * opts.recencyWeight
+          : 0;
+        
+        // Calculate popularity boost (0-1 range)
+        const popularityBoost = calculatePopularityBoost(item);
+        
+        // Apply job-specific boost if available
+        const job = jobs[item.category];
+        let jobBoost = 0;
+        if (job && typeof job.boostCriteria === 'function') {
+          const boost = job.boostCriteria(item, trimmedQuery);
+          if (boost) {
+            jobBoost = boost / 100; // Normalize boost to 0-1
+          }
+        }
+        
+        // Combine scores using weighted average
+        // BM25 and vector are weighted, boosts are additive
+        const hybridScore = 
+          (normalizedBm25Score * opts.bm25Weight) +
+          (vectorSimilarity * opts.vectorWeight) +
+          recencyBoost +
+          popularityBoost +
+          jobBoost;
+        
+        return {
+          ...result,
+          score: hybridScore * 100, // Scale back to 0-100 for consistency
+          // Store component scores for debugging (optional, can be removed in production)
+          _hybridScores: {
+            bm25: normalizedBm25Score,
+            vector: vectorSimilarity,
+            recency: recencyBoost,
+            popularity: popularityBoost,
+            jobBoost: jobBoost,
+            final: hybridScore,
+          },
+        };
+      });
+      
+      // Sort by hybrid score descending
+      rerankedResults.sort((a, b) => b.score - a.score);
+      
+      // Return top results
+      return rerankedResults.slice(0, opts.finalLimit);
+      
+    } catch (e) {
+      console.warn("[Hybrid Search] Vector reranking failed, using BM25 only:", e);
+      // Fallback to BM25 only
+      return topBm25Results.slice(0, opts.finalLimit);
+    }
+  }
+  
+  // If query is too short for vector search, just return BM25 results
+  return topBm25Results.slice(0, opts.finalLimit);
+}
+
+/**
+ * Enhanced hybrid search that also includes vector-only results not found by BM25
+ */
+export async function hybridSearchWithExpansion(
+  bm25Results: CombinedResult[],
+  query: string,
+  allItems: IndexItem[],
+  options: HybridSearchOptions = {},
+): Promise<CombinedResult[]> {
+  const opts = { ...DEFAULT_OPTIONS, ...options };
+  const trimmedQuery = query.trim().toLowerCase();
+  
+  // First, rerank BM25 results
+  const rerankedBm25 = await hybridSearch(bm25Results, query, options);
+  
+  // If query is too short, skip vector expansion
+  if (trimmedQuery.length <= 2) {
+    return rerankedBm25;
+  }
+  
+  // Get vector search results
+  let vectorResults: VectorSearchResult[] = [];
+  try {
+    vectorResults = await searchVectors(trimmedQuery, opts.bm25TopK);
+  } catch (e) {
+    console.warn("[Hybrid Search] Vector search failed:", e);
+    return rerankedBm25;
+  }
+  
+  // Find vector results that weren't in BM25 results
+  const bm25Ids = new Set(bm25Results.map(r => r.item.id));
+  const vectorOnlyResults: CombinedResult[] = [];
+  
+  const now = Date.now();
+  
+  vectorResults.forEach(v => {
+    if (!bm25Ids.has(v.object.id)) {
+      // This is a semantic match that BM25 missed
+      const item = v.object;
+      
+      // Calculate boosts
+      const recencyBoost = opts.recencyBoost 
+        ? calculateRecencyBoost(item, now) * opts.recencyWeight
+        : 0;
+      const popularityBoost = calculatePopularityBoost(item);
+      
+      // Vector-only results get lower base score but high vector similarity
+      const vectorScore = v.similarity * opts.vectorWeight + recencyBoost + popularityBoost;
+      
+      // Apply job-specific boost if available
+      const job = jobs[item.category];
+      let jobBoost = 0;
+      if (job && typeof job.boostCriteria === 'function') {
+        const boost = job.boostCriteria(item, trimmedQuery);
+        if (boost) {
+          jobBoost = boost / 100; // Normalize boost
+        }
+      }
+      
+      vectorOnlyResults.push({
+        id: item.id,
+        type: "dynamic" as const,
+        score: (vectorScore + jobBoost) * 100,
+        item,
+        _hybridScores: {
+          bm25: 0,
+          vector: v.similarity,
+          recency: recencyBoost,
+          popularity: popularityBoost,
+          final: vectorScore + jobBoost,
+        },
+      });
+    }
+  });
+  
+  // Combine reranked BM25 results with vector-only results
+  const allResults = [...rerankedBm25, ...vectorOnlyResults];
+  
+  // Sort by score and return top results
+  allResults.sort((a, b) => b.score - a.score);
+  
+  return allResults.slice(0, opts.finalLimit);
+}
+

src/plugins/built-in/globalSearch/src/search/searchUtils.ts

@@ -6,32 +6,79 @@ import type { IndexItem } from "../indexing/types";
 import { searchVectors } from "./vector/vectorSearch";
 import type { VectorSearchResult } from "./vector/vectorTypes";
 import { jobs } from "../indexing/jobs";
+import { hybridSearchWithExpansion } from "./hybridSearch";
+
+// Search result cache for better performance
+const searchCache = new Map<string, { results: CombinedResult[]; timestamp: number }>();
+const CACHE_TTL = 1000 * 60 * 5; // 5 minutes
+const MAX_CACHE_SIZE = 100;
+
+function getCachedResults(query: string): CombinedResult[] | null {
+  const cached = searchCache.get(query);
+  if (cached && Date.now() - cached.timestamp < CACHE_TTL) {
+    return cached.results;
+  }
+  return null;
+}
+
+function setCachedResults(query: string, results: CombinedResult[]) {
+  // Limit cache size
+  if (searchCache.size >= MAX_CACHE_SIZE) {
+    const firstKey = searchCache.keys().next().value;
+    searchCache.delete(firstKey);
+  }
+  searchCache.set(query, { results, timestamp: Date.now() });
+}
+
+/**
+ * Clears the search result cache
+ */
+export function clearSearchCache(): void {
+  searchCache.clear();
+  console.debug("[Search] Search result cache cleared");
+}
+
+// Listen for cache clear events (e.g., on extension update)
+if (typeof window !== 'undefined') {
+  window.addEventListener('betterseqta-clear-search-cache', () => {
+    clearSearchCache();
+  });
+}
 
 export function createSearchIndexes() {
   const commands = getStaticCommands();
   const dynamicItems = getDynamicItems();
 
+  // Optimized command search options
   const commandOptions = {
     keys: ["text", "category", "keywords"],
     includeScore: true,
     includeMatches: true,
-    threshold: 0.4,
+    threshold: 0.35, // Slightly more permissive for better recall
     minMatchCharLength: 2,
     useExtendedSearch: false,
+    ignoreLocation: false,
+    findAllMatches: false, // Performance optimization
   };
 
+  // Optimized dynamic content search options
   const dynamicOptions = {
     keys: [
-      { name: "text", weight: 2 },
+      { name: "text", weight: 3 }, // Increased weight for title matches
       { name: "content", weight: 1 },
-      { name: "category", weight: 1 },
+      { name: "category", weight: 0.5 }, // Lower weight for category
+      { name: "metadata.subjectName", weight: 1.5 }, // Boost subject name matches
+      { name: "metadata.subjectCode", weight: 1.5 }, // Boost subject code matches
     ],
     includeScore: true,
     includeMatches: true,
-    threshold: 0.4,
+    threshold: 0.5, // More permissive for better partial word matching (increased from 0.4)
-    minMatchCharLength: 2,
+    minMatchCharLength: 2, // Minimum 2 characters for Fuse.js matches (substring fallback handles shorter queries)
-    distance: 100,
+    distance: 100, // Increased to allow matches across longer strings
     useExtendedSearch: true,
+    ignoreLocation: true, // Allow matches anywhere in the string for better partial word matching
+    findAllMatches: true, // Enable to find all matches for better partial word support
+    shouldSort: true,
   };
 
   return {
@@ -105,18 +152,64 @@ export function searchDynamicItems(
   }
 
   const now = Date.now();
-  const searchResults = dynamicContentFuse.search(query, { limit });
+  const queryLower = query.toLowerCase();
+  const queryTrimmed = query.trim();
   
-  return searchResults.map((result: FuseResult<IndexItem>) => {
+  // For short queries (3 chars or less), use a more permissive approach
+  const isShortQuery = queryTrimmed.length <= 3;
+  const searchLimit = Math.min(limit * 3, 50);
+  
+  // First, try Fuse.js search
+  const searchResults = dynamicContentFuse.search(query, { limit: searchLimit });
+  
+  // For short queries, always do a simple substring match to supplement Fuse.js results
+  // This ensures we catch partial word matches like "SAT" in "SAT 1: Differential Calculus"
+  let additionalMatches: IndexItem[] = [];
+  if (isShortQuery) {
+    // Always do substring search for short queries to catch partial word matches
+    for (const item of dynamicIdToItemMap.values()) {
+      const textLower = item.text.toLowerCase();
+      const contentLower = (item.content || '').toLowerCase();
+      const subjectNameLower = (item.metadata?.subjectName || '').toLowerCase();
+      const subjectCodeLower = (item.metadata?.subjectCode || '').toLowerCase();
+      
+      // Check if query appears anywhere in the text, content, or metadata
+      if (textLower.includes(queryLower) || 
+          contentLower.includes(queryLower) ||
+          subjectNameLower.includes(queryLower) ||
+          subjectCodeLower.includes(queryLower)) {
+        // Only add if not already in Fuse.js results
+        if (!searchResults.find(r => r.item.id === item.id)) {
+          additionalMatches.push(item);
+        }
+      }
+    }
+  }
+
+  const results = searchResults.map((result: FuseResult<IndexItem>) => {
     const item = result.item;
     const fuseScore = 10 * (1 - (result.score || 0.5));
     
     let score = fuseScore;
 
+    // Recency boost
     const ageInDays = (now - item.dateAdded) / (1000 * 60 * 60 * 24);
     const recencyBoost = sortByRecent ? 1 / (ageInDays + 1) : 0;
     score += recencyBoost;
     
+    // Boost for exact text matches (especially at the start)
+    const textLower = item.text.toLowerCase();
+    if (textLower.startsWith(queryLower)) {
+      score += 5; // Strong boost for prefix matches
+    } else if (textLower.includes(queryLower)) {
+      score += 2; // Boost for substring matches
+    }
+    
+    // Boost for category matches
+    if (item.category.toLowerCase().includes(queryLower)) {
+      score += 1;
+    }
+
     return {
       id: item.id,
       type: "dynamic" as const,
@@ -125,60 +218,124 @@ export function searchDynamicItems(
       matches: result.matches,
     };
   });
+  
+  // Add additional matches from simple substring search
+  additionalMatches.forEach((item) => {
+    // Check if already in results
+    if (!results.find(r => r.id === item.id)) {
+      const textLower = item.text.toLowerCase();
+      let score = 5; // Base score for substring matches
+      
+      // Boost for prefix matches
+      if (textLower.startsWith(queryLower)) {
+        score += 5;
+      }
+      
+      // Recency boost
+      const ageInDays = (now - item.dateAdded) / (1000 * 60 * 60 * 24);
+      const recencyBoost = sortByRecent ? 1 / (ageInDays + 1) : 0;
+      score += recencyBoost;
+      
+      results.push({
+        id: item.id,
+        type: "dynamic" as const,
+        score,
+        item,
+      });
+    }
+  });
+  
+  // Sort by score and return top results
+  return results.sort((a, b) => b.score - a.score).slice(0, limit);
 }
 
 export async function performSearch(
   query: string,
   commandsFuse: Fuse<StaticCommandItem>,
   commandIdToItemMap: Map<string, StaticCommandItem>,
+  dynamicContentFuse?: Fuse<IndexItem>,
+  dynamicIdToItemMap?: Map<string, IndexItem>,
+  sortByRecent: boolean = true,
 ): Promise<CombinedResult[]> {
-  // Get all results first
+  const trimmedQuery = query.trim().toLowerCase();
+  
+  // Check cache first
+  if (trimmedQuery.length > 2) {
+    const cached = getCachedResults(trimmedQuery);
+    if (cached) {
+      return cached;
+    }
+  }
+
+  // Step 1: Get command results (these don't need hybrid search)
   const commandResults = searchCommands(
     commandsFuse,
-    query,
+    trimmedQuery,
     commandIdToItemMap,
   );
 
-  // Get vector results in parallel
+  // Step 2: Get BM25 results for dynamic items
-  let vectorResults: VectorSearchResult[] = [];
+  let dynamicResults: CombinedResult[] = [];
-  try {
+  if (dynamicContentFuse && dynamicIdToItemMap) {
-    vectorResults = await searchVectors(query);
+    // Get BM25 results first (fast text-based search)
-  } catch (e) {}
+    const bm25Results = searchDynamicItems(
+      dynamicContentFuse,
+      trimmedQuery,
+      dynamicIdToItemMap,
+      50, // Get top 50 for reranking
+      sortByRecent,
+    );
 
-  // Create a map to store our final results, using ID as key to avoid duplicates
+    // Step 3: Apply hybrid search (BM25 + Vector reranking + boosting)
-  const resultMap = new Map<string, CombinedResult>();
+    if (trimmedQuery.length > 2 && bm25Results.length > 0) {
+      try {
+        // Get all items for expansion
+        const allItems = Array.from(dynamicIdToItemMap.values());
         
-  // Add command results first (they keep their original scores)
+        // Apply hybrid search with expansion
-  commandResults.forEach((r) => resultMap.set(r.id, r));
+        dynamicResults = await hybridSearchWithExpansion(
-
+          bm25Results,
-  // Process dynamic results and vector results together
+          trimmedQuery,
-  const seenIds = new Set<string>();
+          allItems,
-
+          {
-  vectorResults.forEach((v) => {
+            bm25TopK: 50,
-    const id = v.object.id;
+            finalLimit: 20, // Return top 20 after reranking
-
+            recencyBoost: sortByRecent,
-    if (!seenIds.has(id)) {
+            bm25Weight: 0.4, // 40% BM25, 60% vector
-      // This is a semantic match that Fuse missed - add it with the vector similarity as score
+            vectorWeight: 0.6,
-      let score = v.similarity * 0.5; // High base score for semantic matches
+            recencyWeight: 0.1,
-      const job = jobs[v.object.category];
+          },
-      if (job && typeof job.boostCriteria === 'function') {
+        );
-        const boost = job.boostCriteria(v.object, query);
+      } catch (e) {
-        if (boost) {
+        console.warn("[Search] Hybrid search failed, using BM25 only:", e);
-          score += boost;
+        // Fallback to BM25 only
-        }
+        dynamicResults = bm25Results.slice(0, 20);
       }
-      resultMap.set(id, {
+    } else {
-        id,
+      // For very short queries or no BM25 results, use BM25 only
-        type: "dynamic" as const,
+      dynamicResults = bm25Results.slice(0, 20);
-        score,
-        item: v.object,
-      });
     }
+  }
+
+  // Step 4: Combine command and dynamic results
+  const allResults = [...commandResults, ...dynamicResults];
+  
+  // Sort by score (commands typically have higher priority)
+  allResults.sort((a, b) => {
+    // Commands always come first if scores are similar
+    if (a.type === "command" && b.type === "dynamic") {
+      return b.score - a.score - 10; // Commands get +10 boost
+    }
+    if (a.type === "dynamic" && b.type === "command") {
+      return b.score - a.score + 10; // Commands get +10 boost
+    }
+    return b.score - a.score;
   });
 
-  // Convert to array and sort by score
+  // Cache results for queries longer than 2 chars
-  const results = Array.from(resultMap.values());
+  if (trimmedQuery.length > 2) {
-  results.sort((a, b) => b.score - a.score);
+    setCachedResults(trimmedQuery, allResults);
+  }
 
-  return results;
+  return allResults;
 }

src/plugins/built-in/globalSearch/src/search/vector/vectorSearch.ts

@@ -1,16 +1,36 @@
 import { EmbeddingIndex, getEmbedding, initializeModel } from "embeddia";
 import type { IndexItem } from "../../indexing/types";
 import type { SearchResult } from "embeddia";
+import { isVectorSearchSupported } from "../../utils/browserDetection";
 
 let vectorIndex: EmbeddingIndex | null = null;
+let initializationAttempted = false;
+let initializationFailed = false;
 
 export async function initVectorSearch() {
+  // Skip initialization if already attempted and failed, or if not supported
+  if (initializationFailed || !isVectorSearchSupported()) {
+    if (!isVectorSearchSupported()) {
+      console.debug("[Vector Search] Vector search not supported in Firefox - using text search only");
+    }
+    return;
+  }
+
+  if (initializationAttempted) {
+    return;
+  }
+
+  initializationAttempted = true;
+
   try {
     await initializeModel();
     vectorIndex = new EmbeddingIndex([]);
     vectorIndex.preloadIndexedDB();
+    console.debug("[Vector Search] Initialized successfully");
   } catch (e) {
-    console.error("Error initializing vector search", e);
+    console.warn("[Vector Search] Failed to initialize vector search (will use text search only):", e);
+    initializationFailed = true;
+    vectorIndex = null;
   }
 }
 
@@ -18,28 +38,111 @@ export interface VectorSearchResult extends SearchResult {
   object: IndexItem & { embedding: number[] };
 }
 
+// Cache for query embeddings to avoid recomputing
+const embeddingCache = new Map<string, number[]>();
+const EMBEDDING_CACHE_TTL = 1000 * 60 * 30; // 30 minutes
+const MAX_EMBEDDING_CACHE_SIZE = 50;
+
+function getCachedEmbedding(query: string): number[] | null {
+  const cached = embeddingCache.get(query);
+  if (cached) {
+    return cached;
+  }
+  return null;
+}
+
+function setCachedEmbedding(query: string, embedding: number[]) {
+  // Limit cache size
+  if (embeddingCache.size >= MAX_EMBEDDING_CACHE_SIZE) {
+    const firstKey = embeddingCache.keys().next().value;
+    embeddingCache.delete(firstKey);
+  }
+  embeddingCache.set(query, embedding);
+}
+
+/**
+ * Clears the embedding cache
+ */
+export function clearEmbeddingCache(): void {
+  embeddingCache.clear();
+  console.debug("[Vector Search] Embedding cache cleared");
+}
+
+// Listen for cache clear events (e.g., on extension update)
+if (typeof window !== 'undefined') {
+  window.addEventListener('betterseqta-clear-embedding-cache', () => {
+    clearEmbeddingCache();
+  });
+}
+
 export async function searchVectors(
   query: string,
   topK: number = 20,
 ): Promise<VectorSearchResult[]> {
-  if (!vectorIndex) await initVectorSearch();
+  // Return empty array if vector search is not supported or failed to initialize
+  if (!isVectorSearchSupported() || initializationFailed) {
+    return [];
+  }
 
-  const queryEmbedding = await getEmbedding(query.slice(0, 100));
+  if (!vectorIndex) {
+    await initVectorSearch();
+    if (!vectorIndex) {
+      return [];
+    }
+  }
 
-  const results = await vectorIndex!.search(queryEmbedding, {
+  // Normalize query for caching
-    topK,
+  const normalizedQuery = query.trim().toLowerCase().slice(0, 100);
-    useStorage: "indexedDB",
-    dedupeEntries: true,
-  });
   
-  // filter results with a similarity below 0.81
+  // Check cache first
-  const filteredResults = results.filter((r) => r.similarity > 0.81);
+  let queryEmbedding = getCachedEmbedding(normalizedQuery);
   
-  return filteredResults as VectorSearchResult[];
+  if (!queryEmbedding) {
+    try {
+      queryEmbedding = await getEmbedding(normalizedQuery);
+      setCachedEmbedding(normalizedQuery, queryEmbedding);
+    } catch (e) {
+      console.warn("[Vector Search] Failed to get embedding:", e);
+      return [];
+    }
+  }
+
+  try {
+    const results = await vectorIndex!.search(queryEmbedding, {
+      topK: Math.min(topK * 2, 30), // Get more results, filter later
+      useStorage: "indexedDB",
+      dedupeEntries: true,
+    });
+
+    // Filter results with a similarity below 0.80 (slightly more permissive)
+    // and sort by similarity descending
+    const filteredResults = results
+      .filter((r) => r.similarity > 0.80)
+      .sort((a, b) => b.similarity - a.similarity)
+      .slice(0, topK);
+
+    return filteredResults as VectorSearchResult[];
+  } catch (e) {
+    console.warn("[Vector Search] Search failed:", e);
+    return [];
+  }
 }
 
 export async function refreshVectorCache() {
-  if (!vectorIndex) await initVectorSearch();
+  if (!isVectorSearchSupported() || initializationFailed) {
-  vectorIndex!.clearIndexedDBCache();
+    return;
-  vectorIndex!.preloadIndexedDB();
+  }
+  
+  if (!vectorIndex) {
+    await initVectorSearch();
+  }
+  
+  if (vectorIndex) {
+    try {
+      vectorIndex.clearIndexedDBCache();
+      vectorIndex.preloadIndexedDB();
+    } catch (e) {
+      console.warn("[Vector Search] Failed to refresh cache:", e);
+    }
+  }
 }

src/plugins/built-in/globalSearch/src/utils/browserDetection.ts

@@ -0,0 +1,30 @@
+import browser from "webextension-polyfill";
+
+/**
+ * Detects if the current browser is Firefox
+ */
+export function isFirefox(): boolean {
+  try {
+    // Firefox-specific API
+    if (typeof (browser.runtime as any).getBrowserInfo === "function") {
+      return true;
+    }
+    // Fallback: check user agent
+    if (typeof navigator !== "undefined") {
+      return navigator.userAgent.toLowerCase().includes("firefox");
+    }
+    return false;
+  } catch {
+    // If we can't detect, assume not Firefox (safer for Chrome/Edge)
+    return false;
+  }
+}
+
+/**
+ * Checks if vector search is supported in the current browser
+ * Currently disabled for Firefox due to security restrictions
+ */
+export function isVectorSearchSupported(): boolean {
+  return !isFirefox();
+}
+

src/plugins/built-in/globalSearch/src/utils/versionCheck.ts

@@ -0,0 +1,115 @@
+import browser from "webextension-polyfill";
+
+const VERSION_STORAGE_KEY = "betterseqta-global-search-version";
+const VERSION_CACHE_KEY = "betterseqta-global-search-cache-version";
+
+/**
+ * Gets the current extension version from the manifest
+ */
+export function getCurrentVersion(): string {
+  try {
+    return browser.runtime.getManifest().version;
+  } catch (e) {
+    console.warn("[Version Check] Failed to get manifest version:", e);
+    return "0.0.0";
+  }
+}
+
+/**
+ * Gets the last stored version from localStorage
+ */
+export function getStoredVersion(): string | null {
+  try {
+    return localStorage.getItem(VERSION_STORAGE_KEY);
+  } catch (e) {
+    console.warn("[Version Check] Failed to get stored version:", e);
+    return null;
+  }
+}
+
+/**
+ * Stores the current version in localStorage
+ */
+export function storeVersion(version: string): void {
+  try {
+    localStorage.setItem(VERSION_STORAGE_KEY, version);
+    localStorage.setItem(VERSION_CACHE_KEY, version);
+  } catch (e) {
+    console.warn("[Version Check] Failed to store version:", e);
+  }
+}
+
+/**
+ * Checks if the extension has been updated and clears caches if needed
+ * Returns true if an update was detected
+ */
+export async function checkAndHandleUpdate(): Promise<boolean> {
+  const currentVersion = getCurrentVersion();
+  const storedVersion = getStoredVersion();
+  
+  // If no stored version, this is first run - store current version
+  if (!storedVersion) {
+    console.debug(`[Version Check] First run detected, storing version ${currentVersion}`);
+    storeVersion(currentVersion);
+    return false;
+  }
+  
+  // If versions match, no update
+  if (storedVersion === currentVersion) {
+    return false;
+  }
+  
+  // Version mismatch detected - extension was updated
+  console.log(`[Version Check] Extension updated from ${storedVersion} to ${currentVersion}, clearing caches...`);
+  
+  // Clear all caches
+  await clearAllCaches();
+  
+  // Store new version
+  storeVersion(currentVersion);
+  
+  return true;
+}
+
+/**
+ * Clears all search-related caches
+ */
+export async function clearAllCaches(): Promise<void> {
+  try {
+    // Clear search result cache (in-memory Map)
+    if (typeof window !== 'undefined') {
+      // Dispatch event to clear caches in other modules
+      window.dispatchEvent(new CustomEvent('betterseqta-clear-search-cache'));
+      window.dispatchEvent(new CustomEvent('betterseqta-clear-embedding-cache'));
+    }
+    
+    // Also try to directly clear caches if modules are already loaded
+    // Use setTimeout to avoid blocking and handle CSS preload errors
+    setTimeout(async () => {
+      try {
+        const { clearSearchCache } = await import("../search/searchUtils");
+        clearSearchCache();
+      } catch (e: any) {
+        // Module might not be loaded yet, or CSS preload error - that's okay
+        if (!e?.message?.includes("preload CSS") && !e?.message?.includes("MIME type")) {
+          console.debug("[Version Check] Could not clear search cache:", e);
+        }
+      }
+      
+      try {
+        const { clearEmbeddingCache } = await import("../search/vector/vectorSearch");
+        clearEmbeddingCache();
+      } catch (e: any) {
+        // Module might not be loaded yet, or CSS preload error - that's okay
+        if (!e?.message?.includes("preload CSS") && !e?.message?.includes("MIME type")) {
+          console.debug("[Version Check] Could not clear embedding cache:", e);
+        }
+      }
+    }, 50);
+    
+    console.debug("[Version Check] All caches cleared");
+  } catch (e) {
+    console.error("[Version Check] Error clearing caches:", e);
+  }
+}
+

src/plugins/core/dynamicLoader.ts

@@ -47,7 +47,17 @@ export function createLazyPlugin<T extends PluginSettings = PluginSettings, S =
         
         // Execute the actual plugin's run function
         return await actualPlugin.run(api);
-      } catch (error) {
+      } catch (error: any) {
+        // Handle Firefox MIME type errors gracefully
+        if (error?.message?.includes("MIME type") || error?.message?.includes("NS_ERROR_CORRUPTED_CONTENT")) {
+          console.error(
+            `[BetterSEQTA+] Failed to load plugin "${lazyPlugin.id}" due to Firefox module loading restrictions. ` +
+            `This may be a build configuration issue. Error:`,
+            error
+          );
+          // Don't throw - allow the extension to continue functioning without this plugin
+          return;
+        }
         console.error(`[BetterSEQTA+] Failed to dynamically load plugin "${lazyPlugin.id}":`, error);
         throw error;
       }