Agent based on any IChatClient

The Microsoft Agent Framework supports creating agents for any inference service that provides a Microsoft.Extensions.AI.IChatClient implementation. This means that there is a very broad range of services that can be used to create agents, including open source models that can be run locally.

In this document, we will use Ollama as an example.

Getting Started

Add the required NuGet packages to your project.

dotnet add package Microsoft.Agents.AI --prerelease

You will also need to add the package for the specific IChatClient implementation you want to use. In this example, we will use OllamaSharp.

dotnet add package OllamaSharp

Creating a ChatClientAgent

To create an agent based on the IChatClient interface, you can use the ChatClientAgent class. The ChatClientAgent class takes IChatClient as a constructor parameter.

First, create an OllamaApiClient to access the Ollama service.

using System;
using Microsoft.Agents.AI;
using OllamaSharp;

using OllamaApiClient chatClient = new(new Uri("http://localhost:11434"), "phi3");

The OllamaApiClient implements the IChatClient interface, so you can use it to create a ChatClientAgent.

AIAgent agent = new ChatClientAgent(
    chatClient,
    instructions: "You are good at telling jokes.",
    name: "Joker");

// Invoke the agent and output the text result.
Console.WriteLine(await agent.RunAsync("Tell me a joke about a pirate."));

Using the Agent

The agent is a standard AIAgent and supports all standard agent operations.

See the Agent getting started tutorials for more information on how to run and interact with agents.

The Microsoft Agent Framework supports creating agents for any inference service that provides a chat client implementation compatible with the ChatClientProtocol. This means that there is a very broad range of services that can be used to create agents, including open source models that can be run locally.

Getting Started

Add the required Python packages to your project.

pip install agent-framework

You may also need to add packages for specific chat client implementations you want to use:

# For Azure AI
pip install agent-framework-azure-ai

# For custom implementations
# Install any required dependencies for your custom client

Built-in Chat Clients

The framework provides several built-in chat client implementations:

OpenAI Chat Client

from agent_framework import ChatAgent
from agent_framework.openai import OpenAIChatClient

# Create agent using OpenAI
agent = ChatAgent(
    chat_client=OpenAIChatClient(model_id="gpt-4o"),
    instructions="You are a helpful assistant.",
    name="OpenAI Assistant"
)

Azure OpenAI Chat Client

from agent_framework import ChatAgent
from agent_framework.azure import AzureOpenAIChatClient

# Create agent using Azure OpenAI
agent = ChatAgent(
    chat_client=AzureOpenAIChatClient(
        model_id="gpt-4o",
        endpoint="https://your-resource.openai.azure.com/",
        api_key="your-api-key"
    ),
    instructions="You are a helpful assistant.",
    name="Azure OpenAI Assistant"
)

Azure AI Agent Client

from agent_framework import ChatAgent
from agent_framework.azure import AzureAIAgentClient
from azure.identity.aio import AzureCliCredential

# Create agent using Azure AI
async with AzureCliCredential() as credential:
    agent = ChatAgent(
        chat_client=AzureAIAgentClient(async_credential=credential),
        instructions="You are a helpful assistant.",
        name="Azure AI Assistant"
    )

Using the Agent

The agent supports all standard agent operations.

See the Agent getting started tutorials for more information on how to run and interact with agents.