Skip to content

MonadicSharp.Azure.OpenAI ​

Railway-Oriented Programming wrapper for Azure OpenAI — typed errors, composable pipelines, built-in integration with MonadicSharp.AI.

Install ​

bash
dotnet add package MonadicSharp.Azure.OpenAI

Requires: MonadicSharp.Azure.Core, MonadicSharp.AI, Azure.AI.OpenAI.

Chat completion ​

csharp
using MonadicSharp.Azure.OpenAI;

var client = new MonadicAzureOpenAiClient(endpoint, new DefaultAzureCredential());

Result<ChatCompletion> result = await client.ChatCompletionMonadicAsync(
    deploymentName: "gpt-4o",
    messages: [
        new SystemChatMessage("You are a helpful assistant."),
        new UserChatMessage(userInput)
    ]);

result.Match(
    onSuccess: completion => Console.WriteLine(completion.Content[0].Text),
    onFailure: err        => logger.LogError("OpenAI call failed: {Err}", err));

Streaming ​

csharp
await foreach (var chunk in client.ChatCompletionStreamMonadicAsync("gpt-4o", messages))
{
    chunk.Match(
        onToken: token => Console.Write(token),
        onError: err   => logger.LogWarning("Stream interrupted: {Err}", err));
}

Embeddings ​

csharp
Result<ReadOnlyMemory<float>> embedding = await client.EmbeddingMonadicAsync(
    deploymentName: "text-embedding-3-small",
    input: documentChunk);

embedding.Match(
    onSuccess: vector => vectorDb.StoreAsync(documentId, vector),
    onFailure: err    => logger.LogError("Embedding failed: {Err}", err));

Compose with MonadicSharp.AI retry ​

csharp
using MonadicSharp.AI;

// Retry on rate-limit and timeout — up to 3 attempts with exponential backoff
var result = await RetryResult<ChatCompletion>
    .ExecuteAsync(
        maxAttempts: 3,
        action:      () => client.ChatCompletionMonadicAsync("gpt-4o", messages),
        shouldRetry: err => err is AzureOpenAiError.RateLimit or AzureOpenAiError.Timeout,
        delay:       attempt => TimeSpan.FromSeconds(Math.Pow(2, attempt)));

Trace with AgentResult ​

csharp
var traced = await AgentResult<string>
    .StartAsync("answer-query", userInput)
    .StepAsync("embed",    input => client.EmbeddingMonadicAsync("text-embedding-3-small", input))
    .StepAsync("retrieve", vector => vectorDb.SearchAsync(vector, topK: 5))
    .StepAsync("complete", context => client.ChatCompletionMonadicAsync("gpt-4o",
        BuildMessages(userInput, context)));

Error types ​

ScenarioError type
Rate limit (429)AzureOpenAiError.RateLimit with RetryAfter
Content filteredAzureOpenAiError.ContentFiltered with Reason
Token limit exceededAzureOpenAiError.TokenLimitExceeded with TokensUsed
Model unavailableAzureOpenAiError.ModelUnavailable
TimeoutAzureOpenAiError.Timeout
Auth failureAzureError.Unauthorized

See Error Mapping for the full reference.

Released under the MIT License.