/// <summary>
///     Configuration options for the AI chat service.
/// </summary>
public sealed class AiChatOptions
{
    /// <summary>AI provider type. Default is <see cref = "AiProvider.OpenAiCompatible"/> (works with LM Studio).</summary>
    public AiProvider Provider { get; set; } = AiProvider.OpenAiCompatible;
    /// <summary>API endpoint URL. When <c>null</c>, uses the provider default.</summary>
    public string? Endpoint { get; set; }
    /// <summary>Model name (e.g., "qwen2.5-3b", "llama3.2", "phi-3.5-mini").</summary>
    public string? DefaultModel { get; set; } = "local-model";
    /// <summary>Custom system prompt. When <c>null</c>, consumers provide their own.</summary>
    public string? SystemPrompt { get; set; }
    /// <summary>Maximum characters of context to include in the prompt. Default is 8000 (~2K tokens).</summary>
    public int MaxContextChars { get; set; } = 8000;
    /// <summary>Whether to stream responses token-by-token. Default is <c>true</c>.</summary>
    public bool StreamResponses { get; set; } = true;
    /// <summary>Generation temperature (0.0 = deterministic, 2.0 = very creative). Default is 0.3.</summary>
    public float Temperature { get; set; } = 0.3f;
    /// <summary>Resolved endpoint URL with provider-specific defaults.</summary>
    public string ResolvedEndpoint => Endpoint ?? Provider switch
    {
        AiProvider.Ollama => "http://localhost:11434",
        AiProvider.Onnx => "",
        _ => "http://localhost:1234/v1"
    };
}