How should I handle OpenAI's system-managed code_interpreter tool in my .NET assistant client?

I'm developing a custom assistant using OpenAI's .NET client and following the GitHub example (link) that defines tools like getNameOfFamilyMemberTool, file_search, and code_interpreter.

I configured my tools as follows:

CSHARPCopied
Tools = { getNameOfFamilyMemberTool, new FileSearchToolDefinition(), new CodeInterpreterToolDefinition() }

Here's what I've set up:

1. getNameOfFamilyMemberTool is implemented locally in a switch case.
2. file_search and code_interpreter are system-defined, so I cannot customize them fully.

Problem

When processing updates from the assistant, I encounter a RequiredActionUpdate for code_interpreter that behaves in two ways:

1. Without handling code_interpreter:
   I get an error:

   "Expected tool outputs for call_ids"
   

2. With a dummy output:
   Returning a placeholder avoids the error but produces a non-functional result:

   CSHARPCopied
   else if (requiredActionUpdate.FunctionName == "code_interpreter")
   {
       // Handle code_interpreter with placeholder
       outputsToSubmit.Add(new ToolOutput(requiredActionUpdate.ToolCallId, "Not implemented yet"));
   }

Question

What is the correct way to handle the code_interpreter tool given that its execution is managed internally by OpenAI, so that I avoid errors without bypassing its intended functionality?

Implementation Context

Below is the relevant section of my code handling tool updates:

CSHARPCopied
if (update is RequiredActionUpdate requiredActionUpdate)
{
    if (requiredActionUpdate.FunctionName == "code_interpreter")
    {
        // Handle code_interpreter with a placeholder
        outputsToSubmit.Add(new ToolOutput(requiredActionUpdate.ToolCallId, "Not implemented yet"));
    }
    else
    {
        // Generic tool handling logic
        var result = await _mediator.Send(requiredActionUpdate.GetCommand(), cancellationToken);
        outputsToSubmit.Add(new ToolOutput(requiredActionUpdate.ToolCallId, result?.ToString()));
    }
}

// Submit tool outputs
if (outputsToSubmit.Count > 0)
{
    asyncUpdates = _client.SubmitToolOutputsToRunStreamingAsync(currentRun.ThreadId, currentRun.Id, outputsToSubmit, cancellationToken);
}

I am looking for advice on how to adjust my handling of code_interpreter so that I rely on OpenAI's internal execution instead of returning a dummy output or triggering errors.