Zum Hauptinhalt springen

.NET SDK

The LLMPort.Client package wraps the official OpenAI .NET SDK and adds typed resources for llm.port platform features.

Requirements

  • .NET 8.0 or later
  • OpenAI 2.*

Installation

dotnet add package LLMPort.Client

Quick start

using LLMPort.Client;

var client = new LLMPortClient(
baseUrl: "http://localhost:4000/v1",
apiKey: "your-api-key"
);

// Standard OpenAI chat completion
var chatClient = client.GetChatClient("assistant-default");
var response = await chatClient.CompleteChatAsync("Hello");
Console.WriteLine(response.Value.Content[0].Text);

Resources

Access llm.port extensions through typed resource properties:

// Sessions
var session = await client.Sessions.CreateAsync("proj-1", "assistant-default");
var sessions = await client.Sessions.ListAsync("proj-1");
var messages = await client.Sessions.MessagesAsync(session.Id);

// Projects
var project = await client.Projects.CreateAsync("My Project");
var projects = await client.Projects.ListAsync();

// Tools
var tools = await client.Tools.AvailableAsync();
var policy = await client.Tools.GetPolicyAsync("sess-1");

// Memory
var facts = await client.Memory.ListAsync();
await client.Memory.CreateAsync("User prefers concise answers");

// Attachments
await client.Attachments.UploadAsync("sess-1", fileStream, "report.pdf");
var attachments = await client.Attachments.ListAsync("sess-1");

// Services
var health = await client.Services.HealthAsync();

Architecture

LLMPortClient creates two internal HTTP clients:

  1. An OpenAIClient pointed at the llm.port gateway — handles standard OpenAI operations
  2. An HttpClient for extension endpoints — handles sessions, tools, memory, and other platform APIs

Use client.GetChatClient(model) for OpenAI operations and the resource properties (client.Sessions, client.Tools, etc.) for llm.port extensions.

Diese Dokumentation wurde mit KI-Unterstützung erstellt und kann Ungenauigkeiten enthalten. Bitte prüfen Sie kritische Details vor dem Produktionseinsatz.