Amazon Bedrock Chat SDK (.NET 9 + Semantic Kernel)

Overview

This repository contains a minimal .NET 9 console app that uses the official Microsoft Semantic Kernel with Amazon Bedrock for AI-powered text and streaming chat completion. It registers the amazon.titan-text-premier-v1:0 model and shows both standard and streaming chat loops.

Key Features

• Bedrock chat completion integration via Semantic Kernel service registration
• Maintains conversational ChatHistory
• Standard response and token streaming examples
• Clean DI setup using Host.CreateApplicationBuilder

Project Structure

• Program.cs – Application entry point, service registration, chat loops
• Amazon.Bedrock.Chat.SDK.csproj – .NET 9 project file

Prerequisites

• .NET 9 SDK installed
• AWS account with access to Amazon Bedrock and Titan model
• Valid AWS credentials available to the application (e.g. environment variables, shared credentials file, or configured IAM role)

Configuration

The model is registered in Program.cs:

builder.Services.AddBedrockChatCompletionService("amazon.titan-text-premier-v1:0");

If you need a different model, replace the model ID above with one you are entitled to use.

Ensure AWS credentials are resolvable by the default SDK chain. Typical environment variables:

AWS_ACCESS_KEY_ID=YOUR_KEY
AWS_SECRET_ACCESS_KEY=YOUR_SECRET
AWS_REGION=us-east-1

(Region must match your Bedrock endpoint availability.)

Build & Run

dotnet restore
dotnet build
dotnet run --project AI.Tutorial/Amazon.Bedrock.Chat.SDK.csproj

You will enter messages after the You: prompt. Press Enter on an empty line to exit each loop.

How It Works

1. Host builder creates a DI container.
2. Bedrock chat completion service is registered.
3. A Kernel is created and retrieved from DI.
4. First loop: non-streaming responses using GetChatMessageContentAsync.
5. Second loop: streaming responses using GetStreamingChatMessageContentsAsync.