通过

步骤 2：添加工具

工具允许代理调用自定义函数，例如提取天气数据、查询数据库或调用 API。

将工具定义为具有 [Description] 属性的任何方法：

using System.ComponentModel;

[Description("Get the weather for a given location.")]
static string GetWeather([Description("The location to get the weather for.")] string location)
    => $"The weather in {location} is cloudy with a high of 15°C.";

使用工具创建代理：

using System;
using Azure.AI.OpenAI;
using Azure.Identity;
using Microsoft.Agents.AI;
using Microsoft.Extensions.AI;

var endpoint = Environment.GetEnvironmentVariable("AZURE_OPENAI_ENDPOINT")
    ?? throw new InvalidOperationException("Set AZURE_OPENAI_ENDPOINT");
var deploymentName = Environment.GetEnvironmentVariable("AZURE_OPENAI_DEPLOYMENT_NAME") ?? "gpt-4o-mini";

AIAgent agent = new AzureOpenAIClient(new Uri(endpoint), new AzureCliCredential())
    .GetChatClient(deploymentName)
    .AsAIAgent(instructions: "You are a helpful assistant.", tools: [AIFunctionFactory.Create(GetWeather)]);

在合适的时候，代理会自动调用您的工具。

Console.WriteLine(await agent.RunAsync("What is the weather like in Amsterdam?"));

小窍门

请参阅 完整示例以获取完整可运行文件。

使用 @tool 修饰器定义工具：

# NOTE: approval_mode="never_require" is for sample brevity.
# Use "always_require" in production for user confirmation before tool execution.
@tool(approval_mode="never_require")
def get_weather(
    location: Annotated[str, Field(description="The location to get the weather for.")],
) -> str:
    """Get the weather for a given location."""
    conditions = ["sunny", "cloudy", "rainy", "stormy"]
    return f"The weather in {location} is {conditions[randint(0, 3)]} with a high of {randint(10, 30)}°C."

使用工具创建代理：

agent = client.as_agent(
    name="WeatherAgent",
    instructions="You are a helpful weather agent. Use the get_weather tool to answer questions.",
    tools=get_weather,
)

小窍门

请参阅 完整示例以获取完整可运行文件。

后续步骤

步骤 3：多轮次对话

更深入：

  • Last updated on