Pass the instruction in the first request

[erossini]

I'm creating an application with ChatGPT and I saw your project. My project for university is to create a simple English teacher that speaks with the user. To do that, I understood, I have to pass the instruction to ChatGPT like

var request = new CreateAssistantRequest(Model.GPT4o);
Assistant = await ChatGPTClient.AssistantsEndpoint.CreateAssistantAsync(
    new CreateAssistantRequest(
        name: $"English Teacher",
        instructions: $"You are a personal English Teacher. Help me learn English by interacting with me " + 
                      "based on my questions and interests.",
        model: Model.GPT4o));

Looking at your code, I didn't understand if and where I can add these instructions. Can you point me in the right direction?

I really appreciate any help you can provide.
Enrico

[danielmonettelli]

Hi @erossini, first of all, thank you for your interest in my project. I've updated the project to Chat Completion, including the current model for both requests and responses. To add instructions for the assistant bot, you'll need to do so in the system role's content, located in OpenAIService.cs. For modifying the tokens, you can adjust them in the Usage class found in CompletionResponse.cs I've included the code where you can change these parameters. Let me know if you encounter any errors so I can improve the code. Thanks!

public async Task<string> AskQuestion(string query)
{
    var completion = new CompletionRequest
    {
        Model = "gpt-4o-mini",
        Messages = new List<MessageRequest>
        {
            new MessageRequest
            {
                Role = "system",
                Content = "You are a helpful assistant."
            },
            new MessageRequest
            {
                Role = "user",
                Content = query
            }
        }
    };

    // ...
}

public class CompletionResponse
{
    public string Id { get; set; }

    public List<Choice> Choices { get; set; }

    public Usage Usage { get; set; }
}

public class Usage
{
    public int Prompt_Tokens { get; set; } = 9;

    public int Completion_Tokens { get; set; } = 12;

    public int Total_Tokens { get; set; } = 21;
}

[erossini]

Thank you so much for your code!

[NikhilSingh750]

Hi @erossini, first of all, thank you !

To implement custom instructions like making the assistant an English teacher, you need to modify the system role's content inside OpenAIService.cs.
Specifically, you can update the AskQuestion method to accept an optional instructions parameter and insert it dynamically.

I've prepared a clean update you can apply:

`public async Task AskQuestion(string query, string instructions = null)
{
string systemContent = string.IsNullOrEmpty(instructions)
? "You are a helpful assistant."
: instructions;

var completion = new CompletionRequest
{
    Model = "gpt-4o-mini",
    Messages = new List<MessageRequest>
    {
        new MessageRequest { Role = "system", Content = systemContent },
        new MessageRequest { Role = "user", Content = query }
    }
};

// (Continue as in existing code)

}

// Keep compatibility

public Task AskQuestion(string query) => AskQuestion(query, null);
Update ConversationViewModel.cs: Add a new property to hold custom instructions:[ObservableProperty]
private string instructions;`

Update ConversationViewModel.cs where you call AskQuestion:
Change your AskQuestionAsync method to pass the Instructions value:

await QueryManagerAsync(q => _openAIService.AskQuestion(q, this.Instructions));
(Optional) Update ConversationView.xaml:
Add an input field for custom instructions:

<Entry Grid.Row="1" Placeholder="Enter custom instructions (optional)" Text="{Binding Instructions}" />

Result:
The user can now set custom instructions before asking a question. If no instructions are provided, the app falls back to "You are a helpful assistant.