Примечание.
Для доступа к этой странице требуется авторизация. Вы можете попробовать войти или изменить каталоги.
Для доступа к этой странице требуется авторизация. Вы можете попробовать изменить каталоги.
Средства позволяют агенту вызывать пользовательские функции, такие как получение данных о погоде, запрос базы данных или вызов API.
Определите инструмент как любой метод с атрибутом [Description] :
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.";
Создайте агент с помощью средства:
using System;
using Azure.AI.Projects;
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 AIProjectClient(new Uri(endpoint), new DefaultAzureCredential())
.AsAIAgent(
model: deploymentName,
instructions: "You are a helpful assistant.",
tools: [AIFunctionFactory.Create(GetWeather)]);
Предупреждение
DefaultAzureCredential удобно для разработки, но требует тщательного рассмотрения в рабочей среде. В рабочей среде рекомендуется использовать определенные учетные данные (например, ManagedIdentityCredential), чтобы избежать проблем с задержкой, непреднамеренной проверки данных аутентификации и потенциальных рисков безопасности из-за резервных механизмов.
Агент автоматически вызовет средство при необходимости:
Console.WriteLine(await agent.RunAsync("What is the weather like in Amsterdam?"));
Подсказка
См. здесь полный пример запускаемого приложения.
Задайте инструмент с декоратором @tool :
# 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."
Создайте агент с помощью средства:
agent = Agent(
client=client,
name="WeatherAgent",
instructions="You are a helpful weather agent. Use the get_weather tool to answer questions.",
tools=[get_weather],
)
Подсказка
Подробный пример можно найти в файле для полного выполнения.
Определите инструмент с помощью functool:
import (
"context"
"fmt"
"github.com/microsoft/agent-framework-go/tool"
"github.com/microsoft/agent-framework-go/tool/functool"
)
var weatherTool = functool.MustNew(functool.Config{
Name: "weather",
Description: "Get the current weather for a given location",
}, func(_ context.Context, location string) (string, error) {
return fmt.Sprintf("The weather in %s is cloudy with a high of 15°C.", location), nil
})
Создайте агент с помощью средства:
a := foundryprovider.NewAgent(
endpoint,
token,
foundryprovider.ModelDeployment(model),
foundryprovider.AgentConfig{
Instructions: "You are a helpful assistant",
Config: agent.Config{
Tools: []tool.Tool{weatherTool},
},
},
)
Агент автоматически вызовет средство при необходимости:
resp, err := a.RunText(ctx, "What is the weather like in Amsterdam?").Collect()
fmt.Println(resp, err)
Подсказка
Подробный пример можно найти в файле для полного выполнения.
Дальнейшие шаги
Вернитесь глубже:
- Обзор средств — сведения обо всех доступных типах инструментов
- Инструменты функций — расширенные шаблоны инструментов функций
- Утверждение инструмента — участие человека в процессе вызова инструментов