Merge remote-tracking branch 'origin/fix/chat_filter_clearing' into all-merges

frontend/app/api/queries/useGetConversationsQuery.ts

@@ -4,6 +4,7 @@ import {
   useQueryClient,
 } from "@tanstack/react-query";
 import type { EndpointType } from "@/contexts/chat-context";
+import { useChat } from "@/contexts/chat-context";
 
 export interface RawConversation {
   response_id: string;
@@ -50,6 +51,7 @@ export const useGetConversationsQuery = (
   options?: Omit<UseQueryOptions, "queryKey" | "queryFn">,
 ) => {
   const queryClient = useQueryClient();
+  const { isOnboardingComplete } = useChat();
 
   async function getConversations(context: { signal?: AbortSignal }): Promise<ChatConversation[]> {
     try {
@@ -95,6 +97,11 @@ export const useGetConversationsQuery = (
     }
   }
 
+  // Extract enabled from options and combine with onboarding completion check
+  // Query is only enabled if onboarding is complete AND the caller's enabled condition is met
+  const callerEnabled = options?.enabled ?? true;
+  const enabled = isOnboardingComplete && callerEnabled;
+
   const queryResult = useQuery(
     {
       queryKey: ["conversations", endpoint, refreshTrigger],
@@ -106,6 +113,7 @@ export const useGetConversationsQuery = (
       refetchOnMount: false, // Don't refetch on every mount
       refetchOnWindowFocus: false, // Don't refetch when window regains focus
       ...options,
+      enabled, // Override enabled after spreading options to ensure onboarding check is applied
     },
     queryClient,
   );

frontend/app/api/queries/useGetNudgesQuery.ts

@@ -3,6 +3,8 @@ import {
   useQuery,
   useQueryClient,
 } from "@tanstack/react-query";
+import { useChat } from "@/contexts/chat-context";
+import { useProviderHealthQuery } from "./useProviderHealthQuery";
 
 type Nudge = string;
 
@@ -27,6 +29,13 @@ export const useGetNudgesQuery = (
 ) => {
   const { chatId, filters, limit, scoreThreshold } = params ?? {};
   const queryClient = useQueryClient();
+  const { isOnboardingComplete } = useChat();
+  
+  // Check if LLM provider is healthy
+  // If health data is not available yet, assume healthy (optimistic)
+  // Only disable if health data exists and shows LLM error
+  const { data: health } = useProviderHealthQuery();
+  const isLLMHealthy = health === undefined || (health?.status === "healthy" && !health?.llm_error);
 
   function cancel() {
     queryClient.removeQueries({
@@ -77,6 +86,11 @@ export const useGetNudgesQuery = (
     }
   }
 
+  // Extract enabled from options and combine with onboarding completion and LLM health checks
+  // Query is only enabled if onboarding is complete AND LLM provider is healthy AND the caller's enabled condition is met
+  const callerEnabled = options?.enabled ?? true;
+  const enabled = isOnboardingComplete && isLLMHealthy && callerEnabled;
+
   const queryResult = useQuery(
     {
       queryKey: ["nudges", chatId, filters, limit, scoreThreshold],
@@ -91,6 +105,7 @@ export const useGetNudgesQuery = (
         return Array.isArray(data) && data.length === 0 ? 5000 : false;
       },
       ...options,
+      enabled, // Override enabled after spreading options to ensure onboarding check is applied
     },
     queryClient,
   );

frontend/app/api/queries/useProviderHealthQuery.ts

@@ -5,6 +5,7 @@ import {
 } from "@tanstack/react-query";
 import { useChat } from "@/contexts/chat-context";
 import { useGetSettingsQuery } from "./useGetSettingsQuery";
+import { useGetTasksQuery } from "./useGetTasksQuery";
 
 export interface ProviderHealthDetails {
   llm_model: string;
@@ -40,11 +41,20 @@ export const useProviderHealthQuery = (
 ) => {
   const queryClient = useQueryClient();
 
-  // Get chat error state from context (ChatProvider wraps the entire app in layout.tsx)
+  // Get chat error state and onboarding completion from context (ChatProvider wraps the entire app in layout.tsx)
-  const { hasChatError, setChatError } = useChat();
+  const { hasChatError, setChatError, isOnboardingComplete } = useChat();
 
   const { data: settings = {} } = useGetSettingsQuery();
 
+  // Check if there are any active ingestion tasks
+  const { data: tasks = [] } = useGetTasksQuery();
+  const hasActiveIngestion = tasks.some(
+    (task) =>
+      task.status === "pending" ||
+      task.status === "running" ||
+      task.status === "processing",
+  );
+
   async function checkProviderHealth(): Promise<ProviderHealthResponse> {
     try {
       const url = new URL("/api/provider/health", window.location.origin);
@@ -55,6 +65,7 @@ export const useProviderHealthQuery = (
       }
 
       // Add test_completion query param if specified or if chat error exists
+      // Use the same testCompletion value that's in the queryKey
       const testCompletion = params?.test_completion ?? hasChatError;
       if (testCompletion) {
         url.searchParams.set("test_completion", "true");
@@ -101,7 +112,10 @@ export const useProviderHealthQuery = (
     }
   }
 
-  const queryKey = ["provider", "health", params?.test_completion];
+  // Include hasChatError in queryKey so React Query refetches when it changes
+  // This ensures the health check runs with test_completion=true when chat errors occur
+  const testCompletion = params?.test_completion ?? hasChatError;
+  const queryKey = ["provider", "health", testCompletion, hasChatError];
   const failureCountKey = queryKey.join("-");
 
   const queryResult = useQuery(
@@ -143,7 +157,11 @@ export const useProviderHealthQuery = (
       refetchOnWindowFocus: false, // Disabled to reduce unnecessary calls on tab switches
       refetchOnMount: true,
       staleTime: 30000, // Consider data stale after 30 seconds
-      enabled: !!settings?.edited && options?.enabled !== false, // Only run after onboarding is complete
+      enabled:
+        !!settings?.edited &&
+        isOnboardingComplete &&
+        !hasActiveIngestion && // Disable health checks when ingestion is happening
+        options?.enabled !== false, // Only run after onboarding is complete
       ...options,
     },
     queryClient,

frontend/app/chat/page.tsx

@@ -10,6 +10,7 @@ import { useTask } from "@/contexts/task-context";
 import { useChatStreaming } from "@/hooks/useChatStreaming";
 import { FILE_CONFIRMATION, FILES_REGEX } from "@/lib/constants";
 import { useLoadingStore } from "@/stores/loadingStore";
+import { useGetConversationsQuery } from "../api/queries/useGetConversationsQuery";
 import { useGetNudgesQuery } from "../api/queries/useGetNudgesQuery";
 import { AssistantMessage } from "./_components/assistant-message";
 import { ChatInput, type ChatInputHandle } from "./_components/chat-input";
@@ -36,6 +37,7 @@ function ChatPage() {
 		forkFromResponse,
 		refreshConversations,
 		refreshConversationsSilent,
+		refreshTrigger,
 		previousResponseIds,
 		setPreviousResponseIds,
 		placeholderConversation,
@@ -71,6 +73,14 @@ function ChatPage() {
 	const lastLoadedConversationRef = useRef<string | null>(null);
 	const { addTask } = useTask();
 
+  console.log(endpoint, refreshTrigger);
+
+	// Check if chat history is loading
+	const { isLoading: isConversationsLoading } = useGetConversationsQuery(
+		endpoint,
+		refreshTrigger,
+	);
+
 	// Use conversation-specific filter instead of global filter
 	const selectedFilter = conversationFilter;
 
@@ -116,7 +126,12 @@ function ChatPage() {
 				if (conversationFilter && typeof window !== "undefined") {
 					const newKey = `conversation_filter_${responseId}`;
 					localStorage.setItem(newKey, conversationFilter.id);
-          console.log("[CHAT] Saved filter association:", newKey, "=", conversationFilter.id);
+					console.log(
+						"[CHAT] Saved filter association:",
+						newKey,
+						"=",
+						conversationFilter.id,
+					);
 				}
 			}
 		},
@@ -507,6 +522,9 @@ function ChatPage() {
 			setTimeout(() => {
 				chatInputRef.current?.focusInput();
 			}, 100);
+		} else if (!conversationData) {
+			// No conversation selected (new conversation)
+			lastLoadedConversationRef.current = null;
 		}
 	}, [
 		conversationData,
@@ -677,7 +695,7 @@ function ChatPage() {
 			scoreThreshold: parsedFilterData?.scoreThreshold ?? 0,
 		},
 		{
-      enabled: isOnboardingComplete, // Only fetch nudges after onboarding is complete
+			enabled: isOnboardingComplete && !isConversationsLoading, // Only fetch nudges after onboarding is complete AND chat history is not loading
 		},
 	);
 
@@ -841,7 +859,12 @@ function ChatPage() {
 
 					// Store the response ID if present for this endpoint
 					if (result.response_id) {
-            console.log("[DEBUG] Received response_id:", result.response_id, "currentConversationId:", currentConversationId);
+						console.log(
+							"[DEBUG] Received response_id:",
+							result.response_id,
+							"currentConversationId:",
+							currentConversationId,
+						);
 
 						setPreviousResponseIds((prev) => ({
 							...prev,
@@ -850,11 +873,16 @@ function ChatPage() {
 
 						// If this is a new conversation (no currentConversationId), set it now
 						if (!currentConversationId) {
-              console.log("[DEBUG] Setting currentConversationId to:", result.response_id);
+							console.log(
+								"[DEBUG] Setting currentConversationId to:",
+								result.response_id,
+							);
 							setCurrentConversationId(result.response_id);
 							refreshConversations(true);
 						} else {
-              console.log("[DEBUG] Existing conversation, doing silent refresh");
+							console.log(
+								"[DEBUG] Existing conversation, doing silent refresh",
+							);
 							// For existing conversations, do a silent refresh to keep backend in sync
 							refreshConversationsSilent();
 						}
@@ -863,7 +891,12 @@ function ChatPage() {
 						if (conversationFilter && typeof window !== "undefined") {
 							const newKey = `conversation_filter_${result.response_id}`;
 							localStorage.setItem(newKey, conversationFilter.id);
-              console.log("[DEBUG] Saved filter association:", newKey, "=", conversationFilter.id);
+							console.log(
+								"[DEBUG] Saved filter association:",
+								newKey,
+								"=",
+								conversationFilter.id,
+							);
 						}
 					}
 				} else {

frontend/app/onboarding/_components/ibm-onboarding.tsx

@@ -1,8 +1,8 @@
 import type { Dispatch, SetStateAction } from "react";
 import { useEffect, useState } from "react";
+import IBMLogo from "@/components/icons/ibm-logo";
 import { LabelInput } from "@/components/label-input";
 import { LabelWrapper } from "@/components/label-wrapper";
-import IBMLogo from "@/components/icons/ibm-logo";
 import { Switch } from "@/components/ui/switch";
 import {
 	Tooltip,
@@ -39,14 +39,16 @@ export function IBMOnboarding({
 	hasEnvApiKey?: boolean;
 }) {
 	const [endpoint, setEndpoint] = useState(
-    alreadyConfigured ? "" : (existingEndpoint || "https://us-south.ml.cloud.ibm.com"),
+		alreadyConfigured
+			? ""
+			: existingEndpoint || "https://us-south.ml.cloud.ibm.com",
 	);
 	const [apiKey, setApiKey] = useState("");
 	const [getFromEnv, setGetFromEnv] = useState(
 		hasEnvApiKey && !alreadyConfigured,
 	);
 	const [projectId, setProjectId] = useState(
-    alreadyConfigured ? "" : (existingProjectId || ""),
+		alreadyConfigured ? "" : existingProjectId || "",
 	);
 
 	const options = [
@@ -93,14 +95,12 @@ export function IBMOnboarding({
 	} = useGetIBMModelsQuery(
 		{
 			endpoint: debouncedEndpoint ? debouncedEndpoint : undefined,
-      apiKey: getFromEnv ? "" : (debouncedApiKey ? debouncedApiKey : undefined),
+			apiKey: getFromEnv ? "" : debouncedApiKey ? debouncedApiKey : undefined,
 			projectId: debouncedProjectId ? debouncedProjectId : undefined,
 		},
 		{
 			enabled:
-        !!debouncedEndpoint ||
+				(!!debouncedEndpoint && !!debouncedApiKey && !!debouncedProjectId) ||
-        !!debouncedApiKey ||
-        !!debouncedProjectId ||
 				getFromEnv ||
 				alreadyConfigured,
 		},

frontend/app/onboarding/_components/onboarding-card.tsx

@@ -507,7 +507,7 @@ const OnboardingCard = ({
                     hasEnvApiKey={
                       currentSettings?.providers?.openai?.has_api_key === true
                     }
-                    alreadyConfigured={providerAlreadyConfigured}
+                    alreadyConfigured={providerAlreadyConfigured && modelProvider === "openai"}
                   />
                 </TabsContent>
                 <TabsContent value="watsonx">
@@ -517,7 +517,7 @@ const OnboardingCard = ({
                     setSampleDataset={setSampleDataset}
                     setIsLoadingModels={setIsLoadingModels}
                     isEmbedding={isEmbedding}
-                    alreadyConfigured={providerAlreadyConfigured}
+                    alreadyConfigured={providerAlreadyConfigured && modelProvider === "watsonx"}
                     existingEndpoint={currentSettings?.providers?.watsonx?.endpoint}
                     existingProjectId={currentSettings?.providers?.watsonx?.project_id}
                     hasEnvApiKey={currentSettings?.providers?.watsonx?.has_api_key === true}
@@ -530,7 +530,7 @@ const OnboardingCard = ({
                     setSampleDataset={setSampleDataset}
                     setIsLoadingModels={setIsLoadingModels}
                     isEmbedding={isEmbedding}
-                    alreadyConfigured={providerAlreadyConfigured}
+                    alreadyConfigured={providerAlreadyConfigured && modelProvider === "ollama"}
                     existingEndpoint={currentSettings?.providers?.ollama?.endpoint}
                   />
                 </TabsContent>

frontend/app/onboarding/_components/onboarding-upload.tsx

@@ -21,6 +21,7 @@ const OnboardingUpload = ({ onComplete }: OnboardingUploadProps) => {
   const [isUploading, setIsUploading] = useState(false);
   const [currentStep, setCurrentStep] = useState<number | null>(null);
   const [uploadedFilename, setUploadedFilename] = useState<string | null>(null);
+  const [uploadedTaskId, setUploadedTaskId] = useState<string | null>(null);
   const [shouldCreateFilter, setShouldCreateFilter] = useState(false);
   const [isCreatingFilter, setIsCreatingFilter] = useState(false);
 
@@ -43,23 +44,26 @@ const OnboardingUpload = ({ onComplete }: OnboardingUploadProps) => {
 
   // Monitor tasks and call onComplete when file processing is done
   useEffect(() => {
-    if (currentStep === null || !tasks) {
+    if (currentStep === null || !tasks || !uploadedTaskId) {
       return;
     }
 
-    // Check if there are any active tasks (pending, running, or processing)
+    // Find the task by task ID from the upload response
-    const activeTasks = tasks.find(
+    const matchingTask = tasks.find((task) => task.task_id === uploadedTaskId);
-      (task) =>
-        task.status === "pending" ||
-        task.status === "running" ||
-        task.status === "processing",
-    );
 
-    // If no active tasks and we have more than 1 task (initial + new upload), complete it
+    // If no matching task found, wait for it to appear
-    if (
+    if (!matchingTask) {
-      (!activeTasks || (activeTasks.processed_files ?? 0) > 0) &&
+      return;
-      tasks.length > 1
+    }
-    ) {
+
+    // Check if the matching task is still active (pending, running, or processing)
+    const isTaskActive =
+      matchingTask.status === "pending" ||
+      matchingTask.status === "running" ||
+      matchingTask.status === "processing";
+
+    // If task is completed or has processed files, complete the onboarding step
+    if (!isTaskActive || (matchingTask.processed_files ?? 0) > 0) {
       // Set to final step to show "Done"
       setCurrentStep(STEP_LIST.length);
 
@@ -91,6 +95,7 @@ const OnboardingUpload = ({ onComplete }: OnboardingUploadProps) => {
           icon: "file",
         });
 
+        // Wait for filter creation to complete before proceeding
         createFilterMutation
           .mutateAsync({
             name: displayName,
@@ -114,9 +119,16 @@ const OnboardingUpload = ({ onComplete }: OnboardingUploadProps) => {
           })
           .finally(() => {
             setIsCreatingFilter(false);
-          });
+            // Refetch nudges to get new ones
-      }
+            refetchNudges();
 
+            // Wait a bit before completing (after filter is created)
+            setTimeout(() => {
+              onComplete();
+            }, 1000);
+          });
+      } else {
+        // No filter to create, just complete
         // Refetch nudges to get new ones
         refetchNudges();
 
@@ -125,7 +137,18 @@ const OnboardingUpload = ({ onComplete }: OnboardingUploadProps) => {
           onComplete();
         }, 1000);
       }
-  }, [tasks, currentStep, onComplete, refetchNudges, shouldCreateFilter, uploadedFilename]);
+    }
+  }, [
+    tasks,
+    currentStep,
+    onComplete,
+    refetchNudges,
+    shouldCreateFilter,
+    uploadedFilename,
+    uploadedTaskId,
+    createFilterMutation,
+    isCreatingFilter,
+  ]);
 
   const resetFileInput = () => {
     if (fileInputRef.current) {
@@ -144,6 +167,11 @@ const OnboardingUpload = ({ onComplete }: OnboardingUploadProps) => {
       const result = await uploadFile(file, true, true); // Pass createFilter=true
       console.log("Document upload task started successfully");
 
+      // Store task ID to track the specific upload task
+      if (result.taskId) {
+        setUploadedTaskId(result.taskId);
+      }
+
       // Store filename and createFilter flag in state to create filter after ingestion succeeds
       if (result.createFilter && result.filename) {
         setUploadedFilename(result.filename);
@@ -176,6 +204,7 @@ const OnboardingUpload = ({ onComplete }: OnboardingUploadProps) => {
 
       // Reset on error
       setCurrentStep(null);
+      setUploadedTaskId(null);
     } finally {
       setIsUploading(false);
     }

frontend/components/chat-renderer.tsx

@@ -47,8 +47,7 @@ export function ChatRenderer({
     refreshConversations,
     startNewConversation,
     setConversationFilter,
-    setCurrentConversationId,
+    setOnboardingComplete,
-    setPreviousResponseIds,
   } = useChat();
 
   // Initialize onboarding state based on local storage and settings
@@ -170,12 +169,17 @@ export function ChatRenderer({
         localStorage.removeItem(ONBOARDING_UPLOAD_STEPS_KEY);
       }
 
-      // Clear ALL conversation state so next message starts fresh
+      // Mark onboarding as complete in context
-      await startNewConversation();
+      setOnboardingComplete(true);
 
-      // Store the user document filter as default for new conversations and load it
+      // Store the user document filter as default for new conversations FIRST
+      // This must happen before startNewConversation() so the filter is available
       await storeDefaultFilterForNewConversations(true);
 
+      // Clear ALL conversation state so next message starts fresh
+      // This will pick up the default filter we just set
+      await startNewConversation();
+
       // Clean up onboarding filter IDs now that we've set the default
       if (typeof window !== "undefined") {
         localStorage.removeItem(ONBOARDING_OPENRAG_DOCS_FILTER_ID_KEY);
@@ -202,6 +206,8 @@ export function ChatRenderer({
       localStorage.removeItem(ONBOARDING_CARD_STEPS_KEY);
       localStorage.removeItem(ONBOARDING_UPLOAD_STEPS_KEY);
     }
+    // Mark onboarding as complete in context
+    setOnboardingComplete(true);
     // Store the OpenRAG docs filter as default for new conversations
     storeDefaultFilterForNewConversations(false);
     setShowLayout(true);

frontend/components/provider-health-banner.tsx

@@ -5,8 +5,8 @@ import { useRouter } from "next/navigation";
 import { useProviderHealthQuery } from "@/app/api/queries/useProviderHealthQuery";
 import type { ModelProvider } from "@/app/settings/_helpers/model-helpers";
 import { Banner, BannerIcon, BannerTitle } from "@/components/ui/banner";
-import { cn } from "@/lib/utils";
 import { useChat } from "@/contexts/chat-context";
+import { cn } from "@/lib/utils";
 import { Button } from "./ui/button";
 
 interface ProviderHealthBannerProps {
@@ -73,8 +73,14 @@ export function ProviderHealthBanner({ className }: ProviderHealthBannerProps) {
 		let errorMessage: string;
 
 		if (llmError && embeddingError) {
-      // Both have errors - show combined message
+			// Both have errors - check if they're the same
-      errorMessage = health?.message || "Provider validation failed";
+			if (llmError === embeddingError) {
+				// Same error for both - show once
+				errorMessage = llmError;
+			} else {
+				// Different errors - show both
+				errorMessage = `${llmError}; ${embeddingError}`;
+			}
 			errorProvider = undefined; // Don't link to a specific provider
 		} else if (llmError) {
 			// Only LLM has error

frontend/contexts/chat-context.tsx

@@ -10,6 +10,7 @@ import {
   useRef,
   useState,
 } from "react";
+import { ONBOARDING_STEP_KEY } from "@/lib/constants";
 
 export type EndpointType = "chat" | "langflow";
 
@@ -81,6 +82,8 @@ interface ChatContextType {
   setConversationFilter: (filter: KnowledgeFilter | null, responseId?: string | null) => void;
   hasChatError: boolean;
   setChatError: (hasError: boolean) => void;
+  isOnboardingComplete: boolean;
+  setOnboardingComplete: (complete: boolean) => void;
 }
 
 const ChatContext = createContext<ChatContextType | undefined>(undefined);
@@ -112,6 +115,37 @@ export function ChatProvider({ children }: ChatProviderProps) {
     useState<KnowledgeFilter | null>(null);
   const [hasChatError, setChatError] = useState(false);
   
+  // Check if onboarding is complete (onboarding step key should be null)
+  const [isOnboardingComplete, setIsOnboardingComplete] = useState(() => {
+    if (typeof window === "undefined") return false;
+    return localStorage.getItem(ONBOARDING_STEP_KEY) === null;
+  });
+
+  // Sync onboarding completion state with localStorage
+  useEffect(() => {
+    const checkOnboarding = () => {
+      if (typeof window !== "undefined") {
+        setIsOnboardingComplete(
+          localStorage.getItem(ONBOARDING_STEP_KEY) === null,
+        );
+      }
+    };
+
+    // Check on mount
+    checkOnboarding();
+
+    // Listen for storage events (for cross-tab sync)
+    window.addEventListener("storage", checkOnboarding);
+
+    return () => {
+      window.removeEventListener("storage", checkOnboarding);
+    };
+  }, []);
+
+  const setOnboardingComplete = useCallback((complete: boolean) => {
+    setIsOnboardingComplete(complete);
+  }, []);
+
   // Listen for ingestion failures and set chat error flag
   useEffect(() => {
     const handleIngestionFailed = () => {
@@ -228,6 +262,10 @@ export function ChatProvider({ children }: ChatProviderProps) {
   const startNewConversation = useCallback(async () => {
     console.log("[CONVERSATION] Starting new conversation");
 
+    // Check if there's existing conversation data - if so, this is a manual "new conversation" action
+    // Check state values before clearing them
+    const hasExistingConversation = conversationData !== null || placeholderConversation !== null;
+    
     // Clear current conversation data and reset state
     setCurrentConversationId(null);
     setPreviousResponseIds({ chat: null, langflow: null });
@@ -261,15 +299,22 @@ export function ChatProvider({ children }: ChatProviderProps) {
           setConversationFilterState(null);
         }
       } else {
-        console.log("[CONVERSATION] No default filter set");
+        // No default filter in localStorage
+        if (hasExistingConversation) {
+          // User is manually starting a new conversation - clear the filter
+          console.log("[CONVERSATION] Manual new conversation - clearing filter");
           setConversationFilterState(null);
-      }
         } else {
-      setConversationFilterState(null);
+          // First time after onboarding - preserve existing filter if set
+          // This prevents clearing the filter when startNewConversation is called multiple times during onboarding
+          console.log("[CONVERSATION] No default filter set, preserving existing filter if any");
+          // Don't clear the filter - it may have been set by storeDefaultFilterForNewConversations
+        }
+      }
     }
 
     // Create a temporary placeholder conversation to show in sidebar
-    const placeholderConversation: ConversationData = {
+    const newPlaceholderConversation: ConversationData = {
       response_id: "new-conversation-" + Date.now(),
       title: "New conversation",
       endpoint: endpoint,
@@ -284,10 +329,10 @@ export function ChatProvider({ children }: ChatProviderProps) {
       last_activity: new Date().toISOString(),
     };
 
-    setPlaceholderConversation(placeholderConversation);
+    setPlaceholderConversation(newPlaceholderConversation);
     // Force immediate refresh to ensure sidebar shows correct state
     refreshConversations(true);
-  }, [endpoint, refreshConversations]);
+  }, [endpoint, refreshConversations, conversationData, placeholderConversation]);
 
   const addConversationDoc = useCallback((filename: string) => {
     setConversationDocs((prev) => [
@@ -375,6 +420,8 @@ export function ChatProvider({ children }: ChatProviderProps) {
       setConversationFilter,
       hasChatError,
       setChatError,
+      isOnboardingComplete,
+      setOnboardingComplete,
     }),
     [
       endpoint,
@@ -396,6 +443,8 @@ export function ChatProvider({ children }: ChatProviderProps) {
       conversationFilter,
       setConversationFilter,
       hasChatError,
+      isOnboardingComplete,
+      setOnboardingComplete,
     ],
   );
 

frontend/lib/upload-utils.ts

@@ -12,6 +12,7 @@ export interface UploadFileResult {
   raw: unknown;
   createFilter?: boolean;
   filename?: string;
+  taskId?: string;
 }
 
 export async function duplicateCheck(
@@ -158,6 +159,7 @@ export async function uploadFile(
       (uploadIngestJson as { upload?: { id?: string } }).upload?.id ||
       (uploadIngestJson as { id?: string }).id ||
       (uploadIngestJson as { task_id?: string }).task_id;
+    const taskId = (uploadIngestJson as { task_id?: string }).task_id;
     const filePath =
       (uploadIngestJson as { upload?: { path?: string } }).upload?.path ||
       (uploadIngestJson as { path?: string }).path ||
@@ -197,6 +199,7 @@ export async function uploadFile(
       raw: uploadIngestJson,
       createFilter: shouldCreateFilter,
       filename,
+      taskId,
     };
 
     return result;

src/api/provider_validation.py

@@ -1,5 +1,6 @@
 """Provider validation utilities for testing API keys and models during onboarding."""
 
+import json
 import httpx
 from utils.container_utils import transform_localhost_url
 from utils.logging_config import get_logger
@@ -7,6 +8,106 @@ from utils.logging_config import get_logger
 logger = get_logger(__name__)
 
 
+def _parse_json_error_message(error_text: str) -> str:
+    """Parse JSON error message and extract just the message field."""
+    try:
+        # Try to parse as JSON
+        error_data = json.loads(error_text)
+        
+        if isinstance(error_data, dict):
+            # WatsonX format: {"errors": [{"code": "...", "message": "..."}], ...}
+            if "errors" in error_data and isinstance(error_data["errors"], list):
+                errors = error_data["errors"]
+                if len(errors) > 0 and isinstance(errors[0], dict):
+                    message = errors[0].get("message", "")
+                    if message:
+                        return message
+                    code = errors[0].get("code", "")
+                    if code:
+                        return f"Error: {code}"
+            
+            # OpenAI format: {"error": {"message": "...", "type": "...", "code": "..."}}
+            if "error" in error_data:
+                error_obj = error_data["error"]
+                if isinstance(error_obj, dict):
+                    message = error_obj.get("message", "")
+                    if message:
+                        return message
+            
+            # Direct message field
+            if "message" in error_data:
+                return error_data["message"]
+            
+            # Generic format: {"detail": "..."}
+            if "detail" in error_data:
+                return error_data["detail"]
+    except (json.JSONDecodeError, ValueError, TypeError):
+        pass
+    
+    # Return original text if not JSON or can't parse
+    return error_text
+
+
+def _extract_error_details(response: httpx.Response) -> str:
+    """Extract detailed error message from API response."""
+    try:
+        # Try to parse JSON error response
+        error_data = response.json()
+        
+        # Common error response formats
+        if isinstance(error_data, dict):
+            # WatsonX format: {"errors": [{"code": "...", "message": "..."}], ...}
+            if "errors" in error_data and isinstance(error_data["errors"], list):
+                errors = error_data["errors"]
+                if len(errors) > 0 and isinstance(errors[0], dict):
+                    # Extract just the message from the first error
+                    message = errors[0].get("message", "")
+                    if message:
+                        return message
+                    # Fallback to code if no message
+                    code = errors[0].get("code", "")
+                    if code:
+                        return f"Error: {code}"
+            
+            # OpenAI format: {"error": {"message": "...", "type": "...", "code": "..."}}
+            if "error" in error_data:
+                error_obj = error_data["error"]
+                if isinstance(error_obj, dict):
+                    message = error_obj.get("message", "")
+                    error_type = error_obj.get("type", "")
+                    code = error_obj.get("code", "")
+                    if message:
+                        details = message
+                        if error_type:
+                            details += f" (type: {error_type})"
+                        if code:
+                            details += f" (code: {code})"
+                        return details
+            
+            # Anthropic format: {"error": {"message": "...", "type": "..."}}
+            if "message" in error_data:
+                return error_data["message"]
+            
+            # Generic format: {"message": "..."}
+            if "detail" in error_data:
+                return error_data["detail"]
+        
+        # If JSON parsing worked but no structured error found, try parsing text
+        response_text = response.text[:500]
+        parsed = _parse_json_error_message(response_text)
+        if parsed != response_text:
+            return parsed
+        return response_text
+        
+    except (json.JSONDecodeError, ValueError):
+        # If JSON parsing fails, try parsing the text as JSON string
+        response_text = response.text[:500] if response.text else f"HTTP {response.status_code}"
+        parsed = _parse_json_error_message(response_text)
+        if parsed != response_text:
+            return parsed
+        return response_text
+
+
 async def validate_provider_setup(
     provider: str,
     api_key: str = None,
@@ -30,7 +131,7 @@ async def validate_provider_setup(
                         If False, performs lightweight validation (no credits consumed). Default: False.
 
     Raises:
-        Exception: If validation fails with message "Setup failed, please try again or select a different provider."
+        Exception: If validation fails, raises the original exception with the actual error message.
     """
     provider_lower = provider.lower()
 
@@ -70,7 +171,8 @@ async def validate_provider_setup(
 
     except Exception as e:
         logger.error(f"Validation failed for provider {provider_lower}: {str(e)}")
-        raise Exception("Setup failed, please try again or select a different provider.")
+        # Preserve the original error message instead of replacing it with a generic one
+        raise
 
 
 async def test_lightweight_health(
@@ -155,8 +257,9 @@ async def _test_openai_lightweight_health(api_key: str) -> None:
             )
 
             if response.status_code != 200:
-                logger.error(f"OpenAI lightweight health check failed: {response.status_code}")
+                error_details = _extract_error_details(response)
-                raise Exception(f"OpenAI API key validation failed: {response.status_code}")
+                logger.error(f"OpenAI lightweight health check failed: {response.status_code} - {error_details}")
+                raise Exception(f"OpenAI API key validation failed: {error_details}")
 
             logger.info("OpenAI lightweight health check passed")
 
@@ -225,8 +328,9 @@ async def _test_openai_completion_with_tools(api_key: str, llm_model: str) -> No
                 )
 
             if response.status_code != 200:
-                logger.error(f"OpenAI completion test failed: {response.status_code} - {response.text}")
+                error_details = _extract_error_details(response)
-                raise Exception(f"OpenAI API error: {response.status_code}")
+                logger.error(f"OpenAI completion test failed: {response.status_code} - {error_details}")
+                raise Exception(f"OpenAI API error: {error_details}")
 
             logger.info("OpenAI completion with tool calling test passed")
 
@@ -260,8 +364,9 @@ async def _test_openai_embedding(api_key: str, embedding_model: str) -> None:
             )
 
             if response.status_code != 200:
-                logger.error(f"OpenAI embedding test failed: {response.status_code} - {response.text}")
+                error_details = _extract_error_details(response)
-                raise Exception(f"OpenAI API error: {response.status_code}")
+                logger.error(f"OpenAI embedding test failed: {response.status_code} - {error_details}")
+                raise Exception(f"OpenAI API error: {error_details}")
 
             data = response.json()
             if not data.get("data") or len(data["data"]) == 0:
@@ -300,8 +405,9 @@ async def _test_watsonx_lightweight_health(
             )
 
             if token_response.status_code != 200:
-                logger.error(f"IBM IAM token request failed: {token_response.status_code}")
+                error_details = _extract_error_details(token_response)
-                raise Exception("Failed to authenticate with IBM Watson - invalid API key")
+                logger.error(f"IBM IAM token request failed: {token_response.status_code} - {error_details}")
+                raise Exception(f"Failed to authenticate with IBM Watson: {error_details}")
 
             bearer_token = token_response.json().get("access_token")
             if not bearer_token:
@@ -335,8 +441,9 @@ async def _test_watsonx_completion_with_tools(
             )
 
             if token_response.status_code != 200:
-                logger.error(f"IBM IAM token request failed: {token_response.status_code}")
+                error_details = _extract_error_details(token_response)
-                raise Exception("Failed to authenticate with IBM Watson")
+                logger.error(f"IBM IAM token request failed: {token_response.status_code} - {error_details}")
+                raise Exception(f"Failed to authenticate with IBM Watson: {error_details}")
 
             bearer_token = token_response.json().get("access_token")
             if not bearer_token:
@@ -388,8 +495,11 @@ async def _test_watsonx_completion_with_tools(
             )
 
             if response.status_code != 200:
-                logger.error(f"IBM Watson completion test failed: {response.status_code} - {response.text}")
+                error_details = _extract_error_details(response)
-                raise Exception(f"IBM Watson API error: {response.status_code}")
+                logger.error(f"IBM Watson completion test failed: {response.status_code} - {error_details}")
+                # If error_details is still JSON, parse it to extract just the message
+                parsed_details = _parse_json_error_message(error_details)
+                raise Exception(f"IBM Watson API error: {parsed_details}")
 
             logger.info("IBM Watson completion with tool calling test passed")
 
@@ -398,6 +508,13 @@ async def _test_watsonx_completion_with_tools(
         raise Exception("Request timed out")
     except Exception as e:
         logger.error(f"IBM Watson completion test failed: {str(e)}")
+        # If the error message contains JSON, parse it to extract just the message
+        error_str = str(e)
+        if "IBM Watson API error: " in error_str:
+            json_part = error_str.split("IBM Watson API error: ", 1)[1]
+            parsed_message = _parse_json_error_message(json_part)
+            if parsed_message != json_part:
+                raise Exception(f"IBM Watson API error: {parsed_message}")
         raise
 
 
@@ -419,8 +536,9 @@ async def _test_watsonx_embedding(
             )
 
             if token_response.status_code != 200:
-                logger.error(f"IBM IAM token request failed: {token_response.status_code}")
+                error_details = _extract_error_details(token_response)
-                raise Exception("Failed to authenticate with IBM Watson")
+                logger.error(f"IBM IAM token request failed: {token_response.status_code} - {error_details}")
+                raise Exception(f"Failed to authenticate with IBM Watson: {error_details}")
 
             bearer_token = token_response.json().get("access_token")
             if not bearer_token:
@@ -450,8 +568,11 @@ async def _test_watsonx_embedding(
             )
 
             if response.status_code != 200:
-                logger.error(f"IBM Watson embedding test failed: {response.status_code} - {response.text}")
+                error_details = _extract_error_details(response)
-                raise Exception(f"IBM Watson API error: {response.status_code}")
+                logger.error(f"IBM Watson embedding test failed: {response.status_code} - {error_details}")
+                # If error_details is still JSON, parse it to extract just the message
+                parsed_details = _parse_json_error_message(error_details)
+                raise Exception(f"IBM Watson API error: {parsed_details}")
 
             data = response.json()
             if not data.get("results") or len(data["results"]) == 0:
@@ -464,6 +585,13 @@ async def _test_watsonx_embedding(
         raise Exception("Request timed out")
     except Exception as e:
         logger.error(f"IBM Watson embedding test failed: {str(e)}")
+        # If the error message contains JSON, parse it to extract just the message
+        error_str = str(e)
+        if "IBM Watson API error: " in error_str:
+            json_part = error_str.split("IBM Watson API error: ", 1)[1]
+            parsed_message = _parse_json_error_message(json_part)
+            if parsed_message != json_part:
+                raise Exception(f"IBM Watson API error: {parsed_message}")
         raise
 
 
@@ -483,8 +611,9 @@ async def _test_ollama_lightweight_health(endpoint: str) -> None:
             )
 
             if response.status_code != 200:
-                logger.error(f"Ollama lightweight health check failed: {response.status_code}")
+                error_details = _extract_error_details(response)
-                raise Exception(f"Ollama endpoint not responding: {response.status_code}")
+                logger.error(f"Ollama lightweight health check failed: {response.status_code} - {error_details}")
+                raise Exception(f"Ollama endpoint not responding: {error_details}")
 
             logger.info("Ollama lightweight health check passed")
 
@@ -537,8 +666,9 @@ async def _test_ollama_completion_with_tools(llm_model: str, endpoint: str) -> N
             )
 
             if response.status_code != 200:
-                logger.error(f"Ollama completion test failed: {response.status_code} - {response.text}")
+                error_details = _extract_error_details(response)
-                raise Exception(f"Ollama API error: {response.status_code}")
+                logger.error(f"Ollama completion test failed: {response.status_code} - {error_details}")
+                raise Exception(f"Ollama API error: {error_details}")
 
             logger.info("Ollama completion with tool calling test passed")
 
@@ -569,8 +699,9 @@ async def _test_ollama_embedding(embedding_model: str, endpoint: str) -> None:
             )
 
             if response.status_code != 200:
-                logger.error(f"Ollama embedding test failed: {response.status_code} - {response.text}")
+                error_details = _extract_error_details(response)
-                raise Exception(f"Ollama API error: {response.status_code}")
+                logger.error(f"Ollama embedding test failed: {response.status_code} - {error_details}")
+                raise Exception(f"Ollama API error: {error_details}")
 
             data = response.json()
             if not data.get("embedding"):
@@ -616,8 +747,9 @@ async def _test_anthropic_lightweight_health(api_key: str) -> None:
             )
 
             if response.status_code != 200:
-                logger.error(f"Anthropic lightweight health check failed: {response.status_code}")
+                error_details = _extract_error_details(response)
-                raise Exception(f"Anthropic API key validation failed: {response.status_code}")
+                logger.error(f"Anthropic lightweight health check failed: {response.status_code} - {error_details}")
+                raise Exception(f"Anthropic API key validation failed: {error_details}")
 
             logger.info("Anthropic lightweight health check passed")
 
@@ -672,8 +804,9 @@ async def _test_anthropic_completion_with_tools(api_key: str, llm_model: str) ->
             )
 
             if response.status_code != 200:
-                logger.error(f"Anthropic completion test failed: {response.status_code} - {response.text}")
+                error_details = _extract_error_details(response)
-                raise Exception(f"Anthropic API error: {response.status_code}")
+                logger.error(f"Anthropic completion test failed: {response.status_code} - {error_details}")
+                raise Exception(f"Anthropic API error: {error_details}")
 
             logger.info("Anthropic completion with tool calling test passed")