Create Flip single-modal to multi-modal (Text, Image, Audio)

[goodtocode]

Censic AI: Flip single-modal to multi-modal (Text, Image, Audio)
Example

Domain

`public sealed class MultimodalInput
{
public List Parts { get; init; } = [];
}

public sealed class AssetPart
{
public string Kind { get; init; } = default!; // "text", "image", "audio"
public string? Text { get; init; }
public byte[]? Bytes { get; init; }
public string? MimeType { get; init; }
}
ORpublic sealed class MultimodalInput
{
public string? Text { get; init; }
public byte[]? Image { get; init; }
public string? ImageMimeType { get; init; }
public byte[]? Audio { get; init; }
public string? AudioMimeType { get; init; }

// Optional metadata
public string? UserId { get; init; }
public string? ConversationId { get; init; }

}
`

Application

public sealed record GenerateSpeechCommand(string Text); public sealed record TranscribeAudioQuery(byte[] Audio, string MimeType); public sealed record GenerateImageCommand(string Prompt); public sealed record DescribeImageQuery(byte[] Image, string MimeType); public sealed record EvaluateAssetQuery(byte[] Asset, string MimeType, string Purpose);

Intention Router (Infrastructure)

`
public static IAgentMessage MapToAgentMessage(MultimodalInput input)
{
if (input.Audio is not null)
return new AgentMessage(
new(input.Audio, input.AudioMimeType!)
);

if (input.Image is not null)
    return new AgentMessage<DescribeImageQuery>(
        new(input.Image, input.ImageMimeType!)
    );

if (!string.IsNullOrWhiteSpace(input.Text))
    return new AgentMessage<GenerateSpeechCommand>(
        new(input.Text)
    );

throw new InvalidOperationException("Unsupported multimodal input.");

}
`