fix: Handle AI errors better in builder (no-changelog) (#18406)

packages/@n8n/ai-workflow-builder.ee/src/ai-workflow-builder-agent.service.ts

@@ -88,7 +88,7 @@ export class AiWorkflowBuilderService {
 				},
 			});
 		} catch (error) {
-			const llmError = new LLMServiceError('Failed to setup LLM models', {
+			const llmError = new LLMServiceError('Failed to connect to LLM Provider', {
 				cause: error,
 				tags: {
 					hasClient: !!this.client,

packages/@n8n/ai-workflow-builder.ee/src/test/workflow-builder-agent.test.ts

@@ -0,0 +1,362 @@
+import type { BaseChatModel } from '@langchain/core/language_models/chat_models';
+import type { ToolMessage } from '@langchain/core/messages';
+import { AIMessage, HumanMessage } from '@langchain/core/messages';
+import type { MemorySaver } from '@langchain/langgraph';
+import { GraphRecursionError } from '@langchain/langgraph';
+import type { Logger } from '@n8n/backend-common';
+import { mock } from 'jest-mock-extended';
+import type { INodeTypeDescription } from 'n8n-workflow';
+import { ApplicationError } from 'n8n-workflow';
+
+import { MAX_AI_BUILDER_PROMPT_LENGTH } from '@/constants';
+import { ValidationError } from '@/errors';
+import type { StreamOutput } from '@/types/streaming';
+import { createStreamProcessor, formatMessages } from '@/utils/stream-processor';
+import {
+	WorkflowBuilderAgent,
+	type WorkflowBuilderAgentConfig,
+	type ChatPayload,
+} from '@/workflow-builder-agent';
+
+jest.mock('@/tools/add-node.tool', () => ({
+	createAddNodeTool: jest.fn().mockReturnValue({ name: 'add_node' }),
+}));
+jest.mock('@/tools/connect-nodes.tool', () => ({
+	createConnectNodesTool: jest.fn().mockReturnValue({ name: 'connect_nodes' }),
+}));
+jest.mock('@/tools/node-details.tool', () => ({
+	createNodeDetailsTool: jest.fn().mockReturnValue({ name: 'node_details' }),
+}));
+jest.mock('@/tools/node-search.tool', () => ({
+	createNodeSearchTool: jest.fn().mockReturnValue({ name: 'node_search' }),
+}));
+jest.mock('@/tools/remove-node.tool', () => ({
+	createRemoveNodeTool: jest.fn().mockReturnValue({ name: 'remove_node' }),
+}));
+jest.mock('@/tools/update-node-parameters.tool', () => ({
+	createUpdateNodeParametersTool: jest.fn().mockReturnValue({ name: 'update_node_parameters' }),
+}));
+jest.mock('@/tools/prompts/main-agent.prompt', () => ({
+	mainAgentPrompt: {
+		invoke: jest.fn().mockResolvedValue('mocked prompt'),
+	},
+}));
+jest.mock('@/utils/operations-processor', () => ({
+	processOperations: jest.fn(),
+}));
+
+jest.mock('@/utils/stream-processor', () => ({
+	createStreamProcessor: jest.fn(),
+	formatMessages: jest.fn(),
+}));
+jest.mock('@/utils/tool-executor', () => ({
+	executeToolsInParallel: jest.fn(),
+}));
+jest.mock('@/chains/conversation-compact', () => ({
+	conversationCompactChain: jest.fn(),
+}));
+
+const mockRandomUUID = jest.fn();
+Object.defineProperty(global, 'crypto', {
+	value: {
+		randomUUID: mockRandomUUID,
+	},
+	writable: true,
+});
+
+describe('WorkflowBuilderAgent', () => {
+	let agent: WorkflowBuilderAgent;
+	let mockLlmSimple: BaseChatModel;
+	let mockLlmComplex: BaseChatModel;
+	let mockLogger: Logger;
+	let mockCheckpointer: MemorySaver;
+	let parsedNodeTypes: INodeTypeDescription[];
+	let config: WorkflowBuilderAgentConfig;
+
+	const mockCreateStreamProcessor = createStreamProcessor as jest.MockedFunction<
+		typeof createStreamProcessor
+	>;
+	const mockFormatMessages = formatMessages as jest.MockedFunction<typeof formatMessages>;
+
+	beforeEach(() => {
+		mockLlmSimple = mock<BaseChatModel>({
+			_llmType: jest.fn().mockReturnValue('test-llm'),
+			bindTools: jest.fn().mockReturnThis(),
+			invoke: jest.fn(),
+		});
+
+		mockLlmComplex = mock<BaseChatModel>({
+			_llmType: jest.fn().mockReturnValue('test-llm-complex'),
+			bindTools: jest.fn().mockReturnThis(),
+			invoke: jest.fn(),
+		});
+
+		mockLogger = mock<Logger>({
+			debug: jest.fn(),
+			info: jest.fn(),
+			warn: jest.fn(),
+			error: jest.fn(),
+		});
+
+		mockCheckpointer = mock<MemorySaver>();
+		mockCheckpointer.getTuple = jest.fn();
+		mockCheckpointer.put = jest.fn();
+		mockCheckpointer.list = jest.fn();
+
+		parsedNodeTypes = [
+			{
+				name: 'TestNode',
+				displayName: 'Test Node',
+				description: 'A test node',
+				version: 1,
+				defaults: {},
+				inputs: [],
+				outputs: [],
+				properties: [],
+				group: ['transform'],
+			} as INodeTypeDescription,
+		];
+
+		config = {
+			parsedNodeTypes,
+			llmSimpleTask: mockLlmSimple,
+			llmComplexTask: mockLlmComplex,
+			logger: mockLogger,
+			checkpointer: mockCheckpointer,
+		};
+
+		agent = new WorkflowBuilderAgent(config);
+	});
+
+	describe('generateThreadId', () => {
+		beforeEach(() => {
+			mockRandomUUID.mockReset();
+		});
+
+		it('should generate thread ID with workflowId and userId', () => {
+			const workflowId = 'workflow-123';
+			const userId = 'user-456';
+			const threadId = WorkflowBuilderAgent.generateThreadId(workflowId, userId);
+			expect(threadId).toBe('workflow-workflow-123-user-user-456');
+		});
+
+		it('should generate thread ID with workflowId but without userId', () => {
+			const workflowId = 'workflow-123';
+			const threadId = WorkflowBuilderAgent.generateThreadId(workflowId);
+			expect(threadId).toMatch(/^workflow-workflow-123-user-\d+$/);
+		});
+
+		it('should generate random UUID when no workflowId provided', () => {
+			const mockUuid = 'test-uuid-1234-5678-9012';
+			mockRandomUUID.mockReturnValue(mockUuid);
+
+			const threadId = WorkflowBuilderAgent.generateThreadId();
+
+			expect(mockRandomUUID).toHaveBeenCalled();
+			expect(threadId).toBe(mockUuid);
+		});
+	});
+
+	describe('chat method', () => {
+		let mockPayload: ChatPayload;
+
+		beforeEach(() => {
+			mockPayload = {
+				message: 'Create a workflow',
+				workflowContext: {
+					currentWorkflow: { id: 'workflow-123' },
+				},
+			};
+		});
+
+		it('should throw ValidationError when message exceeds maximum length', async () => {
+			const longMessage = 'x'.repeat(MAX_AI_BUILDER_PROMPT_LENGTH + 1);
+			const payload: ChatPayload = {
+				message: longMessage,
+			};
+
+			await expect(async () => {
+				const generator = agent.chat(payload);
+				await generator.next();
+			}).rejects.toThrow(ValidationError);
+
+			expect(mockLogger.warn).toHaveBeenCalledWith('Message exceeds maximum length', {
+				messageLength: longMessage.length,
+				maxLength: MAX_AI_BUILDER_PROMPT_LENGTH,
+			});
+		});
+
+		it('should handle valid message length', async () => {
+			const validMessage = 'Create a simple workflow';
+			const payload: ChatPayload = {
+				message: validMessage,
+			};
+
+			// Mock the stream processing to return a proper StreamOutput
+			const mockStreamOutput: StreamOutput = {
+				messages: [
+					{
+						role: 'assistant',
+						type: 'message',
+						text: 'Processing...',
+					},
+				],
+			};
+			const mockAsyncGenerator = (async function* () {
+				yield mockStreamOutput;
+			})();
+
+			mockCreateStreamProcessor.mockReturnValue(mockAsyncGenerator);
+
+			// Mock the LLM to return a simple response
+			(mockLlmSimple.invoke as jest.Mock).mockResolvedValue({
+				content: 'Mocked response',
+				tool_calls: [],
+			});
+
+			const generator = agent.chat(payload);
+			const result = await generator.next();
+
+			expect(result.value).toEqual(mockStreamOutput);
+		});
+
+		it('should handle GraphRecursionError', async () => {
+			mockCreateStreamProcessor.mockImplementation(() => {
+				// eslint-disable-next-line require-yield
+				return (async function* () {
+					throw new GraphRecursionError('Recursion limit exceeded');
+				})();
+			});
+
+			await expect(async () => {
+				const generator = agent.chat(mockPayload);
+				await generator.next();
+			}).rejects.toThrow(ApplicationError);
+		});
+
+		it('should handle invalid request errors', async () => {
+			const invalidRequestError = Object.assign(new Error('Request failed'), {
+				error: {
+					error: {
+						type: 'invalid_request_error',
+						message: 'Invalid API request',
+					},
+				},
+			});
+
+			(mockLlmSimple.invoke as jest.Mock).mockRejectedValue(invalidRequestError);
+
+			await expect(async () => {
+				const generator = agent.chat(mockPayload);
+				await generator.next();
+			}).rejects.toThrow(ApplicationError);
+		});
+
+		it('should rethrow unknown errors', async () => {
+			const unknownError = new Error('Unknown error');
+
+			// Mock createStreamProcessor to throw an unknown error (not GraphRecursionError or abort)
+			mockCreateStreamProcessor.mockImplementation(() => {
+				// eslint-disable-next-line require-yield
+				return (async function* () {
+					throw unknownError;
+				})();
+			});
+
+			await expect(async () => {
+				const generator = agent.chat(mockPayload);
+				await generator.next();
+			}).rejects.toThrow(unknownError);
+		});
+	});
+
+	describe('getSessions', () => {
+		beforeEach(() => {
+			mockFormatMessages.mockImplementation(
+				(messages: Array<AIMessage | HumanMessage | ToolMessage>) =>
+					messages.map((m) => ({ type: m.constructor.name.toLowerCase(), content: m.content })),
+			);
+		});
+
+		it('should return session for existing workflowId', async () => {
+			const workflowId = 'workflow-123';
+			const userId = 'user-456';
+			const mockCheckpoint = {
+				checkpoint: {
+					channel_values: {
+						messages: [new HumanMessage('Hello'), new AIMessage('Hi there!')],
+					},
+					ts: '2023-12-01T12:00:00Z',
+				},
+			};
+
+			(mockCheckpointer.getTuple as jest.Mock).mockResolvedValue(mockCheckpoint);
+
+			const result = await agent.getSessions(workflowId, userId);
+
+			expect(result.sessions).toHaveLength(1);
+			expect(result.sessions[0]).toMatchObject({
+				sessionId: 'workflow-workflow-123-user-user-456',
+				lastUpdated: '2023-12-01T12:00:00Z',
+			});
+			expect(result.sessions[0].messages).toHaveLength(2);
+		});
+
+		it('should return empty sessions when workflowId is undefined', async () => {
+			const result = await agent.getSessions(undefined);
+
+			expect(result.sessions).toHaveLength(0);
+			expect(mockCheckpointer.getTuple).not.toHaveBeenCalled();
+		});
+
+		it('should return empty sessions when no checkpoint exists', async () => {
+			const workflowId = 'workflow-123';
+			(mockCheckpointer.getTuple as jest.Mock).mockRejectedValue(new Error('Thread not found'));
+
+			const result = await agent.getSessions(workflowId);
+
+			expect(result.sessions).toHaveLength(0);
+			expect(mockLogger.debug).toHaveBeenCalledWith('No session found for workflow:', {
+				workflowId,
+				// eslint-disable-next-line @typescript-eslint/no-unsafe-assignment
+				error: expect.any(Error),
+			});
+		});
+
+		it('should handle checkpoint without messages', async () => {
+			const workflowId = 'workflow-123';
+			const mockCheckpoint = {
+				checkpoint: {
+					channel_values: {},
+					ts: '2023-12-01T12:00:00Z',
+				},
+			};
+
+			(mockCheckpointer.getTuple as jest.Mock).mockResolvedValue(mockCheckpoint);
+
+			const result = await agent.getSessions(workflowId);
+
+			expect(result.sessions).toHaveLength(1);
+			expect(result.sessions[0].messages).toHaveLength(0);
+		});
+
+		it('should handle checkpoint with null messages', async () => {
+			const workflowId = 'workflow-123';
+			const mockCheckpoint = {
+				checkpoint: {
+					channel_values: {
+						messages: null,
+					},
+					ts: '2023-12-01T12:00:00Z',
+				},
+			};
+
+			(mockCheckpointer.getTuple as jest.Mock).mockResolvedValue(mockCheckpoint);
+
+			const result = await agent.getSessions(workflowId);
+
+			expect(result.sessions).toHaveLength(1);
+			expect(result.sessions[0].messages).toHaveLength(0);
+		});
+	});
+});

packages/@n8n/ai-workflow-builder.ee/src/workflow-builder-agent.ts

@@ -3,13 +3,14 @@ import type { ToolMessage } from '@langchain/core/messages';
 import { AIMessage, HumanMessage, RemoveMessage } from '@langchain/core/messages';
 import type { RunnableConfig } from '@langchain/core/runnables';
 import type { LangChainTracer } from '@langchain/core/tracers/tracer_langchain';
-import { StateGraph, MemorySaver, END } from '@langchain/langgraph';
+import { StateGraph, MemorySaver, END, GraphRecursionError } from '@langchain/langgraph';
 import type { Logger } from '@n8n/backend-common';
-import type {
-	INodeTypeDescription,
-	IRunExecutionData,
-	IWorkflowBase,
-	NodeExecutionSchema,
+import {
+	ApplicationError,
+	type INodeTypeDescription,
+	type IRunExecutionData,
+	type IWorkflowBase,
+	type NodeExecutionSchema,
 } from 'n8n-workflow';
 
 import { workflowNameChain } from '@/chains/workflow-name';
@@ -148,7 +149,7 @@ export class WorkflowBuilderAgent {
 		};
 
 		const shouldContinue = ({ messages }: typeof WorkflowState.State) => {
-			const lastMessage = messages[messages.length - 1] as AIMessage;
+			const lastMessage: AIMessage = messages[messages.length - 1];
 
 			if (lastMessage.tool_calls?.length) {
 				return 'tools';
@@ -295,10 +296,26 @@ export class WorkflowBuilderAgent {
 	}
 
 	async *chat(payload: ChatPayload, userId?: string, abortSignal?: AbortSignal) {
-		// Check for the message maximum length
-		if (payload.message.length > MAX_AI_BUILDER_PROMPT_LENGTH) {
+		this.validateMessageLength(payload.message);
+
+		const { agent, threadConfig, streamConfig } = this.setupAgentAndConfigs(
+			payload,
+			userId,
+			abortSignal,
+		);
+
+		try {
+			const stream = await this.createAgentStream(payload, streamConfig, agent);
+			yield* this.processAgentStream(stream, agent, threadConfig);
+		} catch (error: unknown) {
+			this.handleStreamError(error);
+		}
+	}
+
+	private validateMessageLength(message: string): void {
+		if (message.length > MAX_AI_BUILDER_PROMPT_LENGTH) {
 			this.logger?.warn('Message exceeds maximum length', {
-				messageLength: payload.message.length,
+				messageLength: message.length,
 				maxLength: MAX_AI_BUILDER_PROMPT_LENGTH,
 			});
 
@@ -306,7 +323,9 @@ export class WorkflowBuilderAgent {
 				`Message exceeds maximum length of ${MAX_AI_BUILDER_PROMPT_LENGTH} characters`,
 			);
 		}
+	}
 
+	private setupAgentAndConfigs(payload: ChatPayload, userId?: string, abortSignal?: AbortSignal) {
 		const agent = this.createWorkflow().compile({ checkpointer: this.checkpointer });
 		const workflowId = payload.workflowContext?.currentWorkflow?.id;
 		// Generate thread ID from workflowId and userId
@@ -320,12 +339,20 @@ export class WorkflowBuilderAgent {
 		const streamConfig = {
 			...threadConfig,
 			streamMode: ['updates', 'custom'],
-			recursionLimit: 30,
+			recursionLimit: 50,
 			signal: abortSignal,
 			callbacks: this.tracer ? [this.tracer] : undefined,
-		} as RunnableConfig;
+		};
 
-		const stream = await agent.stream(
+		return { agent, threadConfig, streamConfig };
+	}
+
+	private async createAgentStream(
+		payload: ChatPayload,
+		streamConfig: RunnableConfig,
+		agent: ReturnType<ReturnType<typeof this.createWorkflow>['compile']>,
+	) {
+		return await agent.stream(
 			{
 				messages: [new HumanMessage({ content: payload.message })],
 				workflowJSON: this.getDefaultWorkflowJSON(payload),
@@ -334,39 +361,95 @@ export class WorkflowBuilderAgent {
 			},
 			streamConfig,
 		);
+	}
 
+	private handleStreamError(error: unknown): never {
+		const invalidRequestErrorMessage = this.getInvalidRequestError(error);
+		if (invalidRequestErrorMessage) {
+			throw new ValidationError(invalidRequestErrorMessage);
+		}
+
+		throw error;
+	}
+
+	private async *processAgentStream(
+		stream: AsyncGenerator<[string, unknown], void, unknown>,
+		agent: ReturnType<ReturnType<typeof this.createWorkflow>['compile']>,
+		threadConfig: RunnableConfig,
+	) {
 		try {
 			const streamProcessor = createStreamProcessor(stream);
 			for await (const output of streamProcessor) {
 				yield output;
 			}
 		} catch (error) {
-			if (
-				error &&
-				typeof error === 'object' &&
-				'message' in error &&
-				typeof error.message === 'string' &&
-				// This is naive, but it's all we get from LangGraph AbortError
-				['Abort', 'Aborted'].includes(error.message)
-			) {
-				// eslint-disable-next-line @typescript-eslint/no-unsafe-member-access
-				const messages = (await agent.getState(threadConfig)).values.messages as Array<
-					AIMessage | HumanMessage | ToolMessage
-				>;
-
-				// Handle abort errors gracefully
-				const abortedAiMessage = new AIMessage({
-					content: '[Task aborted]',
-					id: crypto.randomUUID(),
-				});
-				// TODO: Should we clear tool calls that are in progress?
-				await agent.updateState(threadConfig, { messages: [...messages, abortedAiMessage] });
-				return;
-			}
-			throw error;
+			await this.handleAgentStreamError(error, agent, threadConfig);
 		}
 	}
 
+	private async handleAgentStreamError(
+		error: unknown,
+		agent: ReturnType<ReturnType<typeof this.createWorkflow>['compile']>,
+		threadConfig: RunnableConfig,
+	): Promise<void> {
+		if (
+			error &&
+			typeof error === 'object' &&
+			'message' in error &&
+			typeof error.message === 'string' &&
+			// This is naive, but it's all we get from LangGraph AbortError
+			['Abort', 'Aborted'].includes(error.message)
+		) {
+			// eslint-disable-next-line @typescript-eslint/no-unsafe-member-access
+			const messages = (await agent.getState(threadConfig)).values.messages as Array<
+				AIMessage | HumanMessage | ToolMessage
+			>;
+
+			// Handle abort errors gracefully
+			const abortedAiMessage = new AIMessage({
+				content: '[Task aborted]',
+				id: crypto.randomUUID(),
+			});
+			// TODO: Should we clear tool calls that are in progress?
+			await agent.updateState(threadConfig, { messages: [...messages, abortedAiMessage] });
+			return;
+		}
+
+		// If it's not an abort error, check for GraphRecursionError
+		if (error instanceof GraphRecursionError) {
+			throw new ApplicationError(
+				'Workflow generation stopped: The AI reached the maximum number of steps while building your workflow. This usually means the workflow design became too complex or got stuck in a loop while trying to create the nodes and connections.',
+			);
+		}
+
+		// Re-throw any other errors
+		throw error;
+	}
+
+	private getInvalidRequestError(error: unknown): string | undefined {
+		if (
+			error instanceof Error &&
+			'error' in error &&
+			typeof error.error === 'object' &&
+			error.error
+		) {
+			const innerError = error.error;
+			if ('error' in innerError && typeof innerError.error === 'object' && innerError.error) {
+				const errorDetails = innerError.error;
+				if (
+					'type' in errorDetails &&
+					errorDetails.type === 'invalid_request_error' &&
+					'message' in errorDetails &&
+					typeof errorDetails.message === 'string'
+				) {
+					return errorDetails.message;
+				}
+			}
+		}
+
+		return undefined;
+	}
+
 	async getSessions(workflowId: string | undefined, userId?: string) {
 		// For now, we'll return the current session if we have a workflowId
 		// MemorySaver doesn't expose a way to list all threads, so we'll need to

packages/frontend/@n8n/design-system/src/components/AskAssistantChat/messages/ErrorMessage.vue

@@ -60,6 +60,7 @@ const { t } = useI18n();
 	font-weight: var(--font-weight-regular);
 	line-height: var(--font-line-height-tight);
 	word-break: break-word;
+	flex-grow: 1;
 }
 
 .retryButton {

packages/frontend/editor-ui/src/composables/useBuilderMessages.test.ts

@@ -1453,6 +1453,244 @@ describe('useBuilderMessages', () => {
 		});
 	});
 
+	describe('error message handling with retry', () => {
+		it('should pass retry function to error messages from processAssistantMessages', () => {
+			const retryFn = vi.fn(async () => {});
+			const currentMessages: ChatUI.AssistantMessage[] = [];
+			const newMessages: ChatRequest.MessageResponse[] = [
+				{
+					type: 'error',
+					role: 'assistant',
+					content: 'Something went wrong',
+				},
+			];
+
+			const result = builderMessages.processAssistantMessages(
+				currentMessages,
+				newMessages,
+				'test-id',
+				retryFn,
+			);
+
+			expect(result.messages).toHaveLength(1);
+			const errorMessage = result.messages[0] as ChatUI.ErrorMessage;
+			expect(errorMessage).toMatchObject({
+				id: 'test-id-0',
+				role: 'assistant',
+				type: 'error',
+				content: 'Something went wrong',
+				read: false,
+			});
+			expect(errorMessage.retry).toBe(retryFn);
+		});
+
+		it('should not pass retry function to non-error messages', () => {
+			const retryFn = vi.fn(async () => {});
+			const currentMessages: ChatUI.AssistantMessage[] = [];
+			const newMessages: ChatRequest.MessageResponse[] = [
+				{
+					type: 'message',
+					role: 'assistant',
+					text: 'This is a normal text message',
+				},
+				{
+					role: 'assistant',
+					type: 'tool',
+					toolName: 'add_nodes',
+					toolCallId: 'call-1',
+					status: 'running',
+					updates: [],
+				},
+			];
+
+			const result = builderMessages.processAssistantMessages(
+				currentMessages,
+				newMessages,
+				'test-id',
+				retryFn,
+			);
+
+			expect(result.messages).toHaveLength(2);
+
+			const textMessage = result.messages[0] as ChatUI.TextMessage;
+			expect(textMessage.type).toBe('text');
+			expect('retry' in textMessage).toBe(false);
+
+			const toolMessage = result.messages[1] as ChatUI.ToolMessage;
+			expect(toolMessage.type).toBe('tool');
+			expect('retry' in toolMessage).toBe(false);
+		});
+
+		it('should clear retry from previous error messages when processing new messages', () => {
+			const oldRetryFn = vi.fn(async () => {});
+			const newRetryFn = vi.fn(async () => {});
+
+			const currentMessages: ChatUI.AssistantMessage[] = [
+				{
+					id: 'error-1',
+					role: 'assistant',
+					type: 'error',
+					content: 'First error',
+					retry: oldRetryFn,
+					read: false,
+				} as ChatUI.ErrorMessage,
+				{
+					id: 'error-2',
+					role: 'assistant',
+					type: 'error',
+					content: 'Second error',
+					retry: oldRetryFn,
+					read: false,
+				} as ChatUI.ErrorMessage,
+			];
+
+			const newMessages: ChatRequest.MessageResponse[] = [
+				{
+					type: 'error',
+					role: 'assistant',
+					content: 'New error',
+				},
+			];
+
+			const result = builderMessages.processAssistantMessages(
+				currentMessages,
+				newMessages,
+				'test-id',
+				newRetryFn,
+			);
+
+			expect(result.messages).toHaveLength(3);
+
+			// First error should have retry removed
+			const firstError = result.messages[0] as ChatUI.ErrorMessage;
+			expect(firstError.content).toBe('First error');
+			expect('retry' in firstError).toBe(false);
+
+			// Second error should have retry removed
+			const secondError = result.messages[1] as ChatUI.ErrorMessage;
+			expect(secondError.content).toBe('Second error');
+			expect('retry' in secondError).toBe(false);
+
+			// New error should have the new retry function
+			const newError = result.messages[2] as ChatUI.ErrorMessage;
+			expect(newError.content).toBe('New error');
+			expect(newError.retry).toBe(newRetryFn);
+		});
+
+		it('should only keep retry on the last error message when multiple errors exist', () => {
+			const retryFn = vi.fn(async () => {});
+			const currentMessages: ChatUI.AssistantMessage[] = [];
+			const newMessages: ChatRequest.MessageResponse[] = [
+				{
+					type: 'error',
+					role: 'assistant',
+					content: 'First error in batch',
+				},
+				{
+					type: 'error',
+					role: 'assistant',
+					content: 'Second error in batch',
+				},
+				{
+					type: 'error',
+					role: 'assistant',
+					content: 'Third error in batch',
+				},
+			];
+
+			const result = builderMessages.processAssistantMessages(
+				currentMessages,
+				newMessages,
+				'test-id',
+				retryFn,
+			);
+
+			expect(result.messages).toHaveLength(3);
+
+			// First error should not have retry
+			const firstError = result.messages[0] as ChatUI.ErrorMessage;
+			expect(firstError.content).toBe('First error in batch');
+			expect('retry' in firstError).toBe(false);
+
+			// Second error should not have retry
+			const secondError = result.messages[1] as ChatUI.ErrorMessage;
+			expect(secondError.content).toBe('Second error in batch');
+			expect('retry' in secondError).toBe(false);
+
+			// Only the last error should have retry
+			const lastError = result.messages[2] as ChatUI.ErrorMessage;
+			expect(lastError.content).toBe('Third error in batch');
+			expect(lastError.retry).toBe(retryFn);
+		});
+
+		it('should handle mixed message types and only affect error messages with retry logic', () => {
+			const retryFn = vi.fn(async () => {});
+			const currentMessages: ChatUI.AssistantMessage[] = [
+				{
+					id: 'msg-1',
+					role: 'assistant',
+					type: 'text',
+					content: 'Normal message',
+					read: false,
+				},
+				{
+					id: 'error-1',
+					role: 'assistant',
+					type: 'error',
+					content: 'Old error',
+					retry: retryFn,
+					read: false,
+				} as ChatUI.ErrorMessage,
+			];
+
+			const newMessages: ChatRequest.MessageResponse[] = [
+				{
+					type: 'message',
+					role: 'assistant',
+					text: 'New text message',
+				},
+				{
+					type: 'error',
+					role: 'assistant',
+					content: 'New error message',
+				},
+			];
+
+			const result = builderMessages.processAssistantMessages(
+				currentMessages,
+				newMessages,
+				'test-id',
+				retryFn,
+			);
+
+			expect(result.messages).toHaveLength(4);
+
+			// Normal text message should be unchanged
+			expect(result.messages[0]).toMatchObject({
+				type: 'text',
+				content: 'Normal message',
+			});
+			expect('retry' in result.messages[0]).toBe(false);
+
+			// Old error should have retry removed
+			const oldError = result.messages[1] as ChatUI.ErrorMessage;
+			expect(oldError.content).toBe('Old error');
+			expect('retry' in oldError).toBe(false);
+
+			// New text message should not have retry
+			expect(result.messages[2]).toMatchObject({
+				type: 'text',
+				content: 'New text message',
+			});
+			expect('retry' in result.messages[2]).toBe(false);
+
+			// Only the new error should have retry
+			const newError = result.messages[3] as ChatUI.ErrorMessage;
+			expect(newError.content).toBe('New error message');
+			expect(newError.retry).toBe(retryFn);
+		});
+	});
+
 	describe('applyRatingLogic', () => {
 		it('should apply rating to the last assistant text message after workflow-updated when no tools are running', () => {
 			const messages: ChatUI.AssistantMessage[] = [

packages/frontend/editor-ui/src/composables/useBuilderMessages.ts

@@ -97,6 +97,7 @@ export function useBuilderMessages() {
 		messages: ChatUI.AssistantMessage[],
 		msg: ChatRequest.MessageResponse,
 		messageId: string,
+		retry?: () => Promise<void>,
 	): boolean {
 		let shouldClearThinking = false;
 
@@ -127,6 +128,7 @@ export function useBuilderMessages() {
 				type: 'error',
 				content: msg.content,
 				read: false,
+				retry,
 			});
 			shouldClearThinking = true;
 		}
@@ -248,6 +250,7 @@ export function useBuilderMessages() {
 		currentMessages: ChatUI.AssistantMessage[],
 		newMessages: ChatRequest.MessageResponse[],
 		baseId: string,
+		retry?: () => Promise<void>,
 	): MessageProcessingResult {
 		const mutableMessages = [...currentMessages];
 		let shouldClearThinking = false;
@@ -255,22 +258,36 @@ export function useBuilderMessages() {
 		newMessages.forEach((msg, index) => {
 			// Generate unique ID for each message in the batch
 			const messageId = `${baseId}-${index}`;
-			const clearThinking = processSingleMessage(mutableMessages, msg, messageId);
+			const clearThinking = processSingleMessage(mutableMessages, msg, messageId, retry);
 			shouldClearThinking = shouldClearThinking || clearThinking;
 		});
 
 		const thinkingMessage = determineThinkingMessage(mutableMessages);
 
 		// Apply rating logic only to messages after workflow-updated
-		const finalMessages = applyRatingLogic(mutableMessages);
+		const messagesWithRatingLogic = applyRatingLogic(mutableMessages);
+
+		// Remove retry from all error messages except the last one
+		const messagesWithRetryLogic = removeRetryFromOldErrorMessages(messagesWithRatingLogic);
 
 		return {
-			messages: finalMessages,
+			messages: messagesWithRetryLogic,
 			thinkingMessage,
 			shouldClearThinking: shouldClearThinking && mutableMessages.length > currentMessages.length,
 		};
 	}
 
+	function removeRetryFromOldErrorMessages(messages: ChatUI.AssistantMessage[]) {
+		// Remove retry from all error messages except the last one
+		return messages.map((message, index) => {
+			if (message.type === 'error' && message.retry && index !== messages.length - 1) {
+				const { retry, ...messageWithoutRetry } = message;
+				return messageWithoutRetry;
+			}
+			return message;
+		});
+	}
+
 	function createUserMessage(content: string, id: string): ChatUI.AssistantMessage {
 		return {
 			id,

packages/frontend/editor-ui/src/stores/builder.store.ts

@@ -286,6 +286,7 @@ export const useBuilderStore = defineStore(STORES.BUILDER, () => {
 						chatMessages.value,
 						response.messages,
 						generateMessageId(),
+						retry,
 					);
 					chatMessages.value = result.messages;