README.md

.NET SDK for Inferable

The Inferable .NET Client is a .NET library that allows you to interact with the Inferable API. This library provides functionality to register your .NET functions, manage services, and handle API communication easily.

Installation

To install the Inferable .NET Client, use the following command in your project:

dotnet add package Inferable

Quick Start

Initializing the Client

To create a new Inferable client, use the InferableClient class:

using Inferable;

var options = new InferableOptions
{
    ApiSecret = "your-api-secret", // Replace with your API secret
    BaseUrl = "https://api.inferable.ai" // Optional, uses default if not provided
};

var client = new InferableClient(options);

If you don't provide an API key or base URL, it will attempt to read them from the following environment variables:

• INFERABLE_API_SECRET
• INFERABLE_API_ENDPOINT

Registering a Function

Register a "sayHello" function. This file will register the function with the control-plane.

public class MyInput
{
    public string Message { get; set; }
}

client.Default.RegisterFunction(new FunctionRegistration<MyInput>
{
    Name = "SayHello",
    Description = "A simple greeting function",
    Func = new Func<MyInput, MyResult>>((input) => {
        // Your code here
    }),
});

_ = client.Default.Start();

👉 The DotNet SDK for Inferable reflects the types from the input class of the function.

Unlike the NodeJs SDK, the Dotnet SDK for Inferable reflects the types from the input struct of the function. It uses the NJsonSchema under the hood to generate JSON schemas from C# types through reflection.

If the input class defines System.Text.Json.Serialization attributes, the SDK will use those in the generated schema. This allows for fine-grained control over the schema generation.

Here's an example to illustrate this:

public struct UserInput
{
  [JsonPropertyName("id")]
  public string Id { get; set; }
  [JsonPropertyName("Name")]
  public string Name { get; set; }
  [
    JsonPropertyName("email"),
    JsonIgnore(Condition = JsonIgnoreCondition.WhenWritingNull)
  ]
  public string Email { get; set; }
}

client.Default.RegisterFunction(new FunctionRegistration<MyInput>
{
    Name = "SayHello",
    Description = "A simple greeting function",
    Func = new Func<UserInput, MyResult>>((input) => {
        // Your code here
    }),
});

In this example, the UserInput class uses System.Text.Json.Serialization attributes to define additional properties for the schema:

• The email field is ignored when writing null.

Triggering a run

The following code will create an Inferable run with the prompt "Say hello to John" and the sayHello function attached.

You can inspect the progress of the run:

• in the playground UI via inf app
• in the CLI via inf runs list

var run = await inferable.CreateRunAsync(new CreateRunInput
{
  InitialPrompt = "Say hello to John",
  // Optional: Explicitly attach the `sayHello` function (All functions attached by default)
  AttachedFunctions = new List<FunctionReference>
  {
    new FunctionReference {
      Function = "SayHello",
      Service = "default"
    }
  },
  // Optional: Define a schema for the result to conform to
  ResultSchema = JsonSchema.FromType<RunOutput>();
  // Optional: Subscribe an Inferable function to receive notifications when the run status changes
  //OnStatusChange = new OnStatusChange<RunOutput>
  //{
  //  Function = OnStatusChangeFunction
  //}
});

Console.WriteLine($"Run started: {run.ID}");

// Wait for the run to complete and log
var result = await run.PollAsync(null);

Console.WriteLine($"Run result: {result}");

Runs can also be triggered via the API, CLI or playground UI.

Documentation

• Inferable documentation contains all the information you need to get started with Inferable.

Support

For support or questions, please create an issue in the repository.

Contributing

Contributions to the Inferable .NET Client are welcome. Please ensure that your code adheres to the existing style and includes appropriate tests.