How to reject the tool calling while using `User-Controlled Tool Execution`

How to reject certain tool calling requests in the process of using User-Controlled Tool Execution.

import lombok.RequiredArgsConstructor;
import org.springframework.ai.chat.client.ChatClient;
import org.springframework.ai.chat.messages.AssistantMessage;
import org.springframework.ai.chat.model.ChatResponse;
import org.springframework.ai.chat.model.Generation;
import org.springframework.ai.chat.prompt.Prompt;
import org.springframework.ai.model.tool.ToolCallingChatOptions;
import org.springframework.ai.model.tool.ToolCallingManager;
import org.springframework.ai.model.tool.ToolExecutionResult;
import org.springframework.ai.tool.ToolCallbackProvider;
import org.springframework.stereotype.Component;

import java.util.Scanner;

@Component
@RequiredArgsConstructor
public class ToolCallingAgent {

    private final ChatClient.Builder builder;
    private final ToolCallbackProvider mcpTools;

    public void run() {
        ToolCallingManager toolCallingManager = ToolCallingManager.builder().build();
        ToolCallingChatOptions chatOptions = ToolCallingChatOptions.builder()
                .temperature(0.1)
                .toolCallbacks(mcpTools.getToolCallbacks())
                .internalToolExecutionEnabled(false)
                .build();
        Prompt prompt = new Prompt("""
                Execute tool calls when necessary. If a response does not contain tool call results, 
                skip this tool call and continue with subsequent tool call tasks.
                """, chatOptions);

        ChatClient chatClient = builder.build();
        ChatResponse chatResponse = chatClient
                .prompt(prompt)
                .user("What directories are under the 'code' directory, and what is the purpose of each directory?")
                .call()
                .chatResponse();
        System.out.print(chatResponse.getResult().getOutput().getText());

        Scanner scanner = new Scanner(System.in);
        while (chatResponse.hasToolCalls()) {
            for (Generation generation : chatResponse.getResults()) {
                for (AssistantMessage.ToolCall toolCall : generation.getOutput().getToolCalls()) {
                    System.out.printf("%s (tool call %s)\n", generation.getOutput().getText(), toolCall.toString());
                }
            }
            System.out.print("Allow to execute tool calling ？(y/n): ");
            String confirm = scanner.nextLine().trim().toLowerCase();
            if ("y".equals(confirm)) {
                ToolExecutionResult toolExecutionResult = toolCallingManager.executeToolCalls(prompt, chatResponse);
                prompt = new Prompt(toolExecutionResult.conversationHistory(), chatOptions);
                chatResponse = chatClient.prompt(prompt).call().chatResponse();
                System.out.println(chatResponse.getResult().getOutput().getText());
            } else {
                // TODO Reject tool calling
                // Expecting some methed like `toolCallingManager.rejectToolCalls(prompt, chatResponse);`
                System.out.println("User cancel the tool calling operation");
            }
        }
        System.out.println(chatResponse.getResult().getOutput().getText());
        System.out.println("Finished");
    }
}

The output logs of LLM are currently similar to the content below: when tool calling is not executed, the system repeatedly requests the execution of the same tool calling.

 (tool call ToolCall[id=019a7c85fcdec2d390a2ba15d21b7534, type=function, name=list_directory, arguments={"path": "/path1"}])
Allow to execute tool calling ？(y/n): y

 (tool call ToolCall[id=019a7c86356f17e95d977e986f462643, type=function, name=list_directory, arguments={"path": "/path2"}])
Allow to execute tool calling ？(y/n): y

 (tool call ToolCall[id=019a7c866b12e22800aa8de2a659a339, type=function, name=list_directory, arguments={"path": "/path3"}])
Allow to execute tool calling ？(y/n): n
User cancel the tool calling operation

 (tool call ToolCall[id=019a7c866b12e22800aa8de2a659a339, type=function, name=list_directory, arguments={"path": "/path3"}])
Allow to execute tool calling ？(y/n): n
User cancel the tool calling operation

 (tool call ToolCall[id=019a7c866b12e22800aa8de2a659a339, type=function, name=list_directory, arguments={"path": "/path3"}])
Allow to execute tool calling ？(y/n): n
User cancel the tool calling operation

Is there a possibility to add a method similar to toolCallingManager.rejectToolCalls(prompt, chatResponse) for rejecting tool calls?

Start with ToolCallingManager and its executeToolCalls method, then read the User-Controlled Tool Execution documentation linked in the issue. Trace how the conversation history is updated after execution and define the rejection path so a declined call does not repeat indefinitely; done means the flow can continue or finish without re-requesting the rejected tool call.