mirror of
https://github.com/microsoft/agent-framework.git
synced 2026-06-16 21:04:09 +08:00
.NET: Add AIContextProvider support (#691)
* Add AIContextProvider support * Address feedback. * Address PR comments. * Switch to valuetask and remove parallel calls for AIContextProvider * Remove Model from ModelInvokingAsync method name * Remove agent thread id again and remove it from context provider interface * Add AIContextProvider serialization support to AgentThread and update sample to show this feature * Address PR comments * Improve memory sample * Update sample comment. * Remove AggregateAIContextProvider for now since it makes too many assumptions. We can include it later as a sample if needed. * Update AIContextProviders to have an Invoked method instead of MessagesAddingAsync. * Remove unused using. * Address PR comments. * Address PR comment. * Update comment. * Update comment * Address PR comments.
This commit is contained in:
westey
committed by
GitHub
Unverified
GitHub
parent
89cb94b5c2
commit
66fe1c957c
+2
-4
@@ -39,8 +39,7 @@ namespace SampleApp
|
||||
List<ChatMessage> responseMessages = CloneAndToUpperCase(messages, this.DisplayName).ToList();
|
||||
|
||||
// Notify the thread of the input and output messages.
|
||||
await NotifyThreadOfNewMessagesAsync(thread, messages, cancellationToken);
|
||||
await NotifyThreadOfNewMessagesAsync(thread, responseMessages, cancellationToken);
|
||||
await NotifyThreadOfNewMessagesAsync(thread, messages.Concat(responseMessages), cancellationToken);
|
||||
|
||||
return new AgentRunResponse
|
||||
{
|
||||
@@ -59,8 +58,7 @@ namespace SampleApp
|
||||
List<ChatMessage> responseMessages = CloneAndToUpperCase(messages, this.DisplayName).ToList();
|
||||
|
||||
// Notify the thread of the input and output messages.
|
||||
await NotifyThreadOfNewMessagesAsync(thread, messages, cancellationToken);
|
||||
await NotifyThreadOfNewMessagesAsync(thread, responseMessages, cancellationToken);
|
||||
await NotifyThreadOfNewMessagesAsync(thread, messages.Concat(responseMessages), cancellationToken);
|
||||
|
||||
foreach (var message in responseMessages)
|
||||
{
|
||||
|
||||
@@ -0,0 +1,22 @@
|
||||
<Project Sdk="Microsoft.NET.Sdk">
|
||||
|
||||
<PropertyGroup>
|
||||
<OutputType>Exe</OutputType>
|
||||
<TargetFramework>net9.0</TargetFramework>
|
||||
|
||||
<Nullable>enable</Nullable>
|
||||
<ImplicitUsings>disable</ImplicitUsings>
|
||||
</PropertyGroup>
|
||||
|
||||
<ItemGroup>
|
||||
<PackageReference Include="Azure.AI.OpenAI" />
|
||||
<PackageReference Include="Azure.Identity" />
|
||||
<PackageReference Include="Microsoft.Extensions.AI.OpenAI" />
|
||||
</ItemGroup>
|
||||
|
||||
<ItemGroup>
|
||||
<ProjectReference Include="..\..\..\..\src\Microsoft.Extensions.AI.Agents.OpenAI\Microsoft.Extensions.AI.Agents.OpenAI.csproj" />
|
||||
<ProjectReference Include="..\..\..\..\src\Microsoft.Extensions.AI.Agents\Microsoft.Extensions.AI.Agents.csproj" />
|
||||
</ItemGroup>
|
||||
|
||||
</Project>
|
||||
@@ -0,0 +1,152 @@
|
||||
// Copyright (c) Microsoft. All rights reserved.
|
||||
|
||||
// This sample shows how to add a basic custom memory component to an agent.
|
||||
// The memory component subscribes to all messages added to the conversation and
|
||||
// extracts the user's name and age if provided.
|
||||
// The component adds a prompt to ask for this information if it is not already known
|
||||
// and provides it to the model before each invocation if known.
|
||||
|
||||
using System;
|
||||
using System.Linq;
|
||||
using System.Text;
|
||||
using System.Text.Json;
|
||||
using System.Threading;
|
||||
using System.Threading.Tasks;
|
||||
using Azure.AI.OpenAI;
|
||||
using Azure.Identity;
|
||||
using Microsoft.Extensions.AI;
|
||||
using Microsoft.Extensions.AI.Agents;
|
||||
using OpenAI;
|
||||
using OpenAI.Chat;
|
||||
using SampleApp;
|
||||
|
||||
var endpoint = Environment.GetEnvironmentVariable("AZURE_OPENAI_ENDPOINT") ?? throw new InvalidOperationException("AZURE_OPENAI_ENDPOINT is not set.");
|
||||
var deploymentName = Environment.GetEnvironmentVariable("AZURE_OPENAI_DEPLOYMENT_NAME") ?? "gpt-4o-mini";
|
||||
|
||||
ChatClient chatClient = new AzureOpenAIClient(
|
||||
new Uri(endpoint),
|
||||
new AzureCliCredential())
|
||||
.GetChatClient(deploymentName);
|
||||
|
||||
// Create the agent and provide a factory to add our custom memory component to
|
||||
// all threads created by the agent. Here each new memory component will have its own
|
||||
// user info object, so each thread will have its own memory.
|
||||
// In real world applications/services, where the user info would be persisted in a database,
|
||||
// and preferably shared between multiple threads used by the same user, ensure that the
|
||||
// factory reads the user id from the current context and scopes the memory component
|
||||
// and its storage to that user id.
|
||||
AIAgent agent = chatClient.CreateAIAgent(new ChatClientAgentOptions()
|
||||
{
|
||||
Instructions = "You are a friendly assistant. Always address the user by their name.",
|
||||
AIContextProviderFactory = () => new SampleApp.UserInfoMemory(chatClient.AsIChatClient())
|
||||
});
|
||||
|
||||
// Create a new thread for the conversation.
|
||||
AgentThread thread = agent.GetNewThread();
|
||||
|
||||
Console.WriteLine(">> Use thread with blank memory\n");
|
||||
|
||||
// Invoke the agent and output the text result.
|
||||
Console.WriteLine(await agent.RunAsync("Hello, what is the square root of 9?", thread));
|
||||
Console.WriteLine(await agent.RunAsync("My name is Ruaidhrí", thread));
|
||||
Console.WriteLine(await agent.RunAsync("I am 20 years old", thread));
|
||||
|
||||
// We can serialize the thread. The serialized state will include the state of the memory component.
|
||||
var threadElement = await thread.SerializeAsync();
|
||||
|
||||
Console.WriteLine("\n>> Use deserialized thread with previously created memories\n");
|
||||
|
||||
// Later we can deserialize the thread and continue the conversation with the previous memory component state.
|
||||
var deserializedThread = await agent.DeserializeThreadAsync(threadElement);
|
||||
Console.WriteLine(await agent.RunAsync("What is my name and age?", deserializedThread));
|
||||
|
||||
Console.WriteLine("\n>> Read memories from memory component\n");
|
||||
|
||||
// It's possible to access the memory component via the thread's AIContextProvider property.
|
||||
var userInfo = ((UserInfoMemory)deserializedThread.AIContextProvider!).UserInfo;
|
||||
|
||||
// Output the user info that was captured by the memory component.
|
||||
Console.WriteLine($"MEMORY - User Name: {userInfo.UserName}");
|
||||
Console.WriteLine($"MEMORY - User Age: {userInfo.UserAge}");
|
||||
|
||||
Console.WriteLine("\n>> Use new thread with previously created memories\n");
|
||||
|
||||
// Create a new thread.
|
||||
thread = agent.GetNewThread();
|
||||
|
||||
// It is also possible to add the memory component to an individual thread only instead of all
|
||||
// threads via the factory above.
|
||||
// In this case we will also use the same user info object, so this thread will share the same
|
||||
// memories as the previous thread.
|
||||
thread.AIContextProvider = new UserInfoMemory(chatClient.AsIChatClient(), userInfo);
|
||||
|
||||
// Invoke the agent and output the text result.
|
||||
// This time the agent should remember the user's name and use it in the response.
|
||||
Console.WriteLine(await agent.RunAsync("What is my name and age?", thread));
|
||||
|
||||
namespace SampleApp
|
||||
{
|
||||
/// <summary>
|
||||
/// Sample memory component that can remember a user's name and age.
|
||||
/// </summary>
|
||||
internal sealed class UserInfoMemory(IChatClient chatClient, UserInfo? userInfo = null) : AIContextProvider
|
||||
{
|
||||
public UserInfo UserInfo { get; set; } = userInfo ?? new();
|
||||
|
||||
public override async ValueTask InvokedAsync(InvokedContext context, CancellationToken cancellationToken = default)
|
||||
{
|
||||
// Try and extract the user name and age from the message if we don't have it already and it's a user message.
|
||||
if ((this.UserInfo.UserName == null || this.UserInfo.UserAge == null) && context.RequestMessages.Any(x => x.Role == ChatRole.User))
|
||||
{
|
||||
var result = await chatClient.GetResponseAsync<UserInfo>(
|
||||
context.RequestMessages,
|
||||
new ChatOptions()
|
||||
{
|
||||
Instructions = "Extract the user's name and age from the message if present. If not present return nulls."
|
||||
},
|
||||
cancellationToken: cancellationToken);
|
||||
|
||||
this.UserInfo.UserName ??= result.Result.UserName;
|
||||
this.UserInfo.UserAge ??= result.Result.UserAge;
|
||||
}
|
||||
}
|
||||
|
||||
public override ValueTask<AIContext> InvokingAsync(InvokingContext context, CancellationToken cancellationToken = default)
|
||||
{
|
||||
StringBuilder instructions = new();
|
||||
|
||||
// If we don't already know the user's name and age, add instructions to ask for them, otherwise just provide what we have to the context.
|
||||
instructions.AppendLine(
|
||||
this.UserInfo.UserName == null ?
|
||||
"Ask the user for their name and politely decline to answer any questions until they provide it." :
|
||||
$"The user's name is {this.UserInfo.UserName}.");
|
||||
|
||||
instructions.AppendLine(
|
||||
this.UserInfo.UserAge == null ?
|
||||
"Ask the user for their age and politely decline to answer any questions until they provide it." :
|
||||
$"The user's age is {this.UserInfo.UserAge}.");
|
||||
|
||||
return new ValueTask<AIContext>(new AIContext
|
||||
{
|
||||
Instructions = instructions.ToString()
|
||||
});
|
||||
}
|
||||
|
||||
public override ValueTask<JsonElement?> SerializeAsync(JsonSerializerOptions? jsonSerializerOptions = null, CancellationToken cancellationToken = default)
|
||||
{
|
||||
return new ValueTask<JsonElement?>(JsonSerializer.SerializeToElement(this.UserInfo, jsonSerializerOptions));
|
||||
}
|
||||
|
||||
public override ValueTask DeserializeAsync(JsonElement serializedState, JsonSerializerOptions? jsonSerializerOptions = null, CancellationToken cancellationToken = default)
|
||||
{
|
||||
this.UserInfo = JsonSerializer.Deserialize<UserInfo>(serializedState, jsonSerializerOptions) ?? new UserInfo();
|
||||
return default;
|
||||
}
|
||||
}
|
||||
|
||||
internal sealed class UserInfo
|
||||
{
|
||||
public string? UserName { get; set; }
|
||||
public int? UserAge { get; set; }
|
||||
}
|
||||
}
|
||||
@@ -37,7 +37,8 @@ Before you begin, ensure you have the following prerequisites:
|
||||
|[Dependency injection with a simple agent](./Agent_Step09_DependencyInjection/)|This sample demonstrates how to add and resolve an agent with a dependency injection container|
|
||||
|[Exposing a simple agent as MCP tool](./Agent_Step10_AsMcpTool/)|This sample demonstrates how to expose an agent as an MCP tool|
|
||||
|[Using images with a simple agent](./Agent_Step11_UsingImages/)|This sample demonstrates how to use image multi-modality with an AI agent|
|
||||
|[Exposing a simple agent a function tool](./Agent_Step12_AsFunctionTool/)|This sample demonstrates how to expose an agent as a function tool|
|
||||
|[Exposing a simple agent as a function tool](./Agent_Step12_AsFunctionTool/)|This sample demonstrates how to expose an agent as a function tool|
|
||||
|[Using memory with an agent](./Agent_Step12_Memory/)|This sample demonstrates how to create a simple memory component and use it with an agent|
|
||||
|
||||
## Running the samples from the console
|
||||
|
||||
|
||||
Reference in New Issue
Block a user