可观测性代理

重要提示

你需要是边境预览计划的一部分，才能提前访问 Microsoft Agent 365。边界将你直接与Microsoft最新的 AI 创新联系起来。边境预览版受客户协议现有预览条款的约束。由于这些功能仍在开发中，其可用性和功能可能会随时间而变化。

若要参与代理 365 生态系统，必须将代理 365 可观测性功能添加到代理中。 Agent 365 Observability 基于 OpenTelemetry （OTel）构建，提供统一的框架，用于在所有代理平台上一致且安全地捕获遥测。通过实现此必需组件，可以让 IT 管理员监视代理在 Microsoft 管理中心（MAC）中的活动，并允许安全团队使用 Defender 和 Purview 进行合规性和威胁检测。

关键优势

端到端可见性：为每个代理调用（包括会话、工具调用和异常）捕获全面的遥测数据，从而提供跨平台的完整可跟踪性。
安全性与符合性启用：将统一审核日志馈送到 Defender 和 Purview 中，为代理启用高级安全方案和合规性报告。
跨平台灵活性：基于 OTel 标准构建，并支持各种运行时和平台，例如 Copilot Studio、Foundry 和未来的代理框架。
管理员的作效率：在 MAC 中提供集中的可观测性，减少故障排除时间，并通过管理代理的 IT 团队的基于角色的访问控制改进治理。

安装

使用这些命令为代理 365 支持的语言安装可观测性模块。

pip install microsoft-agents-a365-observability-core
pip install microsoft-agents-a365-runtime

npm install @microsoft/agents-a365-observability
npm install @microsoft/agents-a365-runtime

dotnet add package Microsoft.Agents.A365.Observability
dotnet add package Microsoft.Agents.A365.Observability.Runtime

配置

可观测性所需的环境变量包括：

环境变量	Description
`ENABLE_OBSERVABILITY=true`	用于启用/禁用跟踪的标志。默认情况下：`false`
`ENABLE_A365_OBSERVABILITY_EXPORTER=true`	`True` 将日志导出到我们的服务。否则回退到控制台导出程序

from microsoft_agents_a365.observability.core import config

def token_resolver(agent_id: str, tenant_id: str) -> str | None:
    # Implement secure token retrieval here
    return "Bearer <token>"

config.configure(
    service_name="my-agent-service",
    service_namespace="my.namespace",
    token_resolver=token_resolver,
)

排除令牌解析程序以登录到控制台。

可观测性所需的环境变量包括：

环境变量	Description
`ENABLE_OBSERVABILITY=true`	用于启用/禁用跟踪的标志。默认情况下：`false`
`ENABLE_A365_OBSERVABILITY_EXPORTER=true`	`True` 将日志导出到我们的服务。否则回退到控制台导出程序

import { ObservabilityManager } from '@microsoft/agents-a365-observability';

const tokenResolver = (agentId, tenantId) => {
    // Your token resolution logic here
    return "your-token";
};

// Advanced configuration with builder pattern
const builder = ObservabilityManager.configure(builder =>
  builder
    .withService('my-agent-service', '1.0.0')
    .withTokenResolver((agentId, tenantId) => {
      return tokenResolver(agentId, tenantId);
    })
);

builder.start();

在 appsettings.json 中将 EnableAgent365Exporter 设置为 true。

在 Program.cs中，添加到 Agent365ExporterOptions 服务集合。此更改配置跟踪导出程序用于检索令牌的委托。

使用 AddA365Tracing(). 添加可观测性相关依赖项。

using Microsoft.Agents.A365.Observability.Runtime;

var builder = WebApplication.CreateBuilder(args);

builder.Services.AddSingleton(sp =>
    {
        return new Agent365ExporterOptions
        {
            ClusterCategory = "prod",
            TokenResolver = async (agentId, tenantId) =>
            {
                // It's recommended to implement caching in your token provider for performance.
                var token = await tokenProvider.GetObservabilityTokenAsync(agentId, tenantId);
                return token;
            }
        };
    });

builder.AddA365Tracing();

行李属性

用于 BaggageBuilder 设置在请求中流经所有范围的上下文信息。 SDK 实现一个 SpanProcessor 副本，将所有无空行李条目复制到新启动的跨度，而无需覆盖现有属性。

from microsoft_agents_a365.observability.core.middleware.baggage_builder import BaggageBuilder

with (
    BaggageBuilder()
    .tenant_id("tenant-123")
    .agent_id("agent-456")
    .correlation_id("corr-789")
    .build()
):
    # Any spans started in this context will receive these as attributes
    pass

import { BaggageBuilder } from '@microsoft/agents-a365-observability';

// Create and apply baggage context
using baggageScope = new BaggageBuilder()
  // Core identifiers
  .tenantId('tenant-123')
  .agentId('agent-456')
  .correlationId('correlation-789')  
  .build();

// Execute operations within the baggage context
baggageScope.run(() => {
  // All spans created within this context will inherit the baggage values
  
  // Invoke another agent
  using agentScope = InvokeAgentScope.start(invokeDetails, agentDetails, tenantDetails);
  // ... agent logic
  
  // Execute tools
  using toolScope = ExecuteToolScope.start(toolDetails, agentDetails, tenantDetails);
  // ... tool logic
});

 using var baggageScope = new BaggageBuilder()
          .TenantId('tenant-123')
          .AgentId('agent-456')
          .CorrelationId('correlation-789')
          .Build();
// Any spans started in this context will receive them as attributes.

令牌解析程序

使用代理 365 导出程序时，必须提供返回身份验证令牌的令牌解析程序函数。将 Agent 365 可观测性 SDK 与代理托管框架配合使用时，可以使用代理活动中的令牌生成令牌TurnContext

from microsoft_agents.activity import load_configuration_from_env
from microsoft_agents.authentication.msal import MsalConnectionManager
from microsoft_agents.hosting.aiohttp import CloudAdapter
from microsoft_agents.hosting.core import (
    AgentApplication,
    Authorization,
    MemoryStorage,
    TurnContext,
    TurnState,
)
from microsoft_agents_a365.runtime.environment_utils import (
    get_observability_authentication_scope,
)

agents_sdk_config = load_configuration_from_env(environ)

STORAGE = MemoryStorage()
CONNECTION_MANAGER = MsalConnectionManager(**agents_sdk_config)
ADAPTER = CloudAdapter(connection_manager=CONNECTION_MANAGER)
ADAPTER.use(TranscriptLoggerMiddleware(ConsoleTranscriptLogger()))
AUTHORIZATION = Authorization(STORAGE, CONNECTION_MANAGER, **agents_sdk_config)

AGENT_APP = AgentApplication[TurnState](
    storage=STORAGE, adapter=ADAPTER, authorization=AUTHORIZATION, **agents_sdk_config
)

@AGENT_APP.activity("message", auth_handlers=["AGENTIC"])
async def on_message(context: TurnContext, _state: TurnState):
    aau_auth_token = await AGENT_APP.auth.exchange_token(
                        context,
                        scopes=get_observability_authentication_scope(),
                        auth_handler_id="AGENTIC",
                    )
    # cache this auth token and return via token resolver

import {
  TurnState,
  AgentApplication,
  MemoryStorage,
  TurnContext,
} from '@microsoft/agents-hosting';
import { Activity, ActivityTypes } from '@microsoft/agents-activity';
import { getObservabilityAuthenticationScope } from '@microsoft/agents-a365-runtime';

interface ConversationState {
  count: number;
}
type ApplicationTurnState = TurnState<ConversationState>;

const downloader = new AttachmentDownloader();

const storage = new MemoryStorage();

export const agentApplication = new AgentApplication<ApplicationTurnState>({
  authorization: {
        agentic: { } // We have the type and scopes set in the .env file
      },
  storage,
});

agentApplication.onActivity(
  ActivityTypes.Message,
  async (context: TurnContext, state: ApplicationTurnState) => {
    const aauAuthToken = await agentApplication.authorization.exchangeToken(context,'agentic', {
        scopes: getObservabilityAuthenticationScope() 
      } )
    // cache this auth token and return via token resolver
  };

将代理令牌解析程序添加到服务集合。

using Microsoft.Agents.A365.Observability;

builder.Services.AddAgenticTracingExporter();

在代理应用程序中，注册令牌。

using Microsoft.Agents.Builder;
using Microsoft.Agents.Core.Models;
using Microsoft.Extensions.Logging;
using Microsoft.Agents.A365.Observability.Caching;
using System;
using System.Threading.Tasks;

public class MyAgent : AgentApplication
{
    private readonly IExporterTokenCache<AgenticTokenStruct> _agentTokenCache;
    private readonly ILogger<MyAgent> _logger;

    public MyAgent(AgentApplicationOptions options, IExporterTokenCache<AgenticTokenStruct> agentTokenCache, ILogger<MyAgent> logger)
        : base(options)
    {
        _agentTokenCache = agentTokenCache ?? throw new ArgumentNullException(nameof(agentTokenCache));
        _logger = logger ?? throw new ArgumentNullException(nameof(logger));
    }

    protected async Task MessageActivityAsync(ITurnContext turnContext, ITurnState turnState, CancellationToken cancellationToken)
    {
        using var baggageScope = new BaggageBuilder()
            .TenantId(turnContext.Activity.Recipient.TenantId)
            .AgentId(turnContext.Activity.Recipient.AgenticAppId)
            .Build();

        try
        {
            _agentTokenCache.RegisterObservability(
                turnContext.Activity.Recipient.AgenticAppId,
                turnContext.Activity.Recipient.TenantId,
                new AgenticTokenStruct
                {
                    UserAuthorization = UserAuthorization,
                    TurnContext = turnContext
                },
                EnvironmentUtils.GetObservabilityAuthenticationScope()
            );
        }
        catch (Exception ex)
        {
            _logger.LogWarning($"Error registering for observability: {ex.Message}");
        }
    }
}

自动检测

自动检测会自动侦听代理框架（SDK）现有的跟踪遥测信号，并将其转发到代理 365 可观测性服务。这样，开发人员就无需手动编写监视代码、简化设置并确保一致的性能跟踪。

跨多个 SDK 和平台支持自动检测：

Platform	支持的 SDK/框架
.NET	语义内核， OpenAI，代理框架
Python	语义内核， OpenAI，代理框架， LangChain
Node.js	OpenAI

备注

对自动检测的支持因平台和 SDK 实现而异。

语义内核

自动检测需要使用行李生成器。使用 BaggageBuilder. 设置代理 ID 和租户 ID。

安装包

pip install microsoft-agents-a365-observability-extensions-semantic-kernel

配置可观测性

from microsoft_agents_a365.observability.core.config import configure
from microsoft_agents_a365.observability.extensions.semantic_kernel import SemanticKernelInstrumentor

# Configure observability
configure(
    service_name="my-semantic-kernel-agent",
    service_namespace="ai.agents"
)

# Enable auto-instrumentation
instrumentor = SemanticKernelInstrumentor()
instrumentor.instrument()

# Your Semantic Kernel code is now automatically traced

将依赖项添加到服务集合。

using Microsoft.Agents.A365.Observability.Extensions.SemanticKernel;

builder.Services.AddTracing(config => config.WithSemanticKernel());

设置 AgentId 和使用 TenantIdBaggageBuilder。确保创建 ChatCompletionAgent 与传递给 BaggageBuilder的代理 ID 匹配时使用的 ID。

using Microsoft.Agents.A365.Observability.Extensions.SemanticKernel;
using Microsoft.Agents.A365.Observability.Runtime.Common;
public class MyAgent
{
    public async Task<AgentResponse> ProcessUserRequest(string userInput)
    {
      using var baggageScope = new BaggageBuilder()
          .AgentId(<your-agent-id>) // NOTE: This will be the agent ID with which the TokenResolver delegate is invoked. 
          .TenantId(<your-tenant-id>) // NOTE: This will be the tenant ID with which the TokenResolver delegate is invoked.
          .Build();
      
      var chatCompletionAgent = new ChatCompletionAgent
      {
          // NOTE: This will be the agent ID with which the TokenResolver delegate is invoked. Should match above.
          Id = <your-agent-id>,
          ...
      };
    }
}

OpenAI

自动检测需要使用行李生成器。使用 BaggageBuilder. 设置代理 ID 和租户 ID。

安装此包。

pip install microsoft-agents-a365-observability-extensions-openai

配置可观测性

from microsoft_agents_a365.observability.core.config import configure
from microsoft_agents_a365.observability.extensions.openai_agents import OpenAIAgentsTraceInstrumentor

# Configure observability
configure(
    service_name="my-openai-agent",
    service_namespace="ai.agents"
)

# Enable auto-instrumentation
instrumentor = OpenAIAgentsTraceInstrumentor()
instrumentor.instrument()

# Your OpenAI Agents code is now automatically traced

安装包

npm install @microsoft/agents-a365-observability-extensions-openai

配置可观测性

import { ObservabilityManager } from '@microsoft/agents-a365-observability';
import { OpenAIAgentsTraceInstrumentor } from '@microsoft/agents-a365-observability-extensions-openai';

// Configure observability first
const sdk = ObservabilityManager.configure((builder) =>
  builder
    .withService('My Agent Service', '1.0.0')
    .withConsoleExporter(true)
);

// Create and enable the instrumentor
const instrumentor = new OpenAIAgentsTraceInstrumentor({
  enabled: true,
  tracerName: 'openai-agents-tracer',
  tracerVersion: '1.0.0'
});

sdk.start();

instrumentor.enable();

将依赖项添加到服务集合。

using Microsoft.Agents.A365.Observability.Extensions.OpenAI;

builder.Services.AddTracing(config => config.WithOpenAI());

设置 AgentId 和使用 TenantIdBaggageBuilder。对于工具调用，请使用 Trace() 实例启动 ChatToolCall 跟踪。

using Microsoft.Agents.A365.Observability.Extensions.OpenAI;
using Microsoft.Agents.A365.Observability.Runtime.Common;

public class MyAgent
{
    public async Task<AgentResponse> ProcessUserRequest(string userInput)
    {
      using var baggageScope = new BaggageBuilder()
          .AgentId(<your-agent-id>) // NOTE: This will be the agent ID with which the TokenResolver delegate is invoked. 
          .TenantId(<your-tenant-id>) // NOTE: This will be the tenant ID with which the TokenResolver delegate is invoked.
          .Build();
      
      // NOTE: This will be the agent and tenant ID with which the TokenResolver delegate will be invoked. 
      using var scope = chatToolCall.Trace(agentId: <your-agent-id>, <your-tenant-id>);
    }
}

代理框架

自动检测需要使用行李生成器。使用 BaggageBuilder 设置代理 ID 和租户 ID。

安装包

pip install microsoft-agents-a365-observability-extensions-agent-framework

配置可观测性

from microsoft_agents_a365.observability.core.config import configure
from microsoft_agents_a365.observability.extensions.agentframework.trace_instrumentor import (
    AgentFrameworkInstrumentor,
)

# Configure observability
configure(
    service_name="AgentFrameworkTracingWithAzureOpenAI",
    service_namespace="AgentFrameworkTesting",
)

# Enable auto-instrumentation
AgentFrameworkInstrumentor().instrument()

将依赖项添加到服务集合。

using Microsoft.Agents.A365.Observability.Extensions.AgentFramework;

builder.Services.AddTracing(config => config.WithAgentFramework());

设置 AgentId 和使用 TenantIdBaggageBuilder。

using Microsoft.Agents.A365.Observability.Runtime.Common;

public class MyAgent : AgentApplication
{
    protected async Task MessageActivityAsync(ITurnContext turnContext, ITurnState turnState, CancellationToken cancellationToken)
    {
      using var baggageScope = new BaggageBuilder()
          .AgentId(<your-agent-id>) // NOTE: This will be the agent ID with which the TokenResolver delegate is invoked. 
          .TenantId(<your-tenant-id>) // NOTE: This will be the tenant ID with which the TokenResolver delegate is invoked.
          .Build();
    }
}

LangChain 框架

自动检测需要使用行李生成器。使用 BaggageBuilder. 设置代理 ID 和租户 ID。

安装此包。

pip install microsoft-agents-a365-observability-extensions-langchain

配置可观测性

from microsoft_agents_a365.observability.core.config import configure
from microsoft_agents_a365.observability.extensions.langchain import CustomLangChainInstrumentor

# Configure observability
configure(
    service_name="my-langchain-agent",
    service_namespace="ai.agents"
)

# Enable auto-instrumentation
CustomLangChainInstrumentor()

# Your LangChain code is now automatically traced

手动检测

代理 365 可观测性 SDK 可用于了解代理的内部工作。 SDK 提供三个可以启动的范围： InvokeAgentScope、 ExecuteToolScope和 InferenceScope。

代理调用

应在代理进程开始时使用此范围。使用调用代理范围，可以捕获正在调用的当前代理、代理用户数据等属性。

from microsoft_agents_a365.observability.core.invoke_agent_scope import InvokeAgentScope
from microsoft_agents_a365.observability.core.invoke_agent_details import InvokeAgentDetails
from microsoft_agents_a365.observability.core.tenant_details import TenantDetails
from microsoft_agents_a365.observability.core.request import Request

invoke_details = InvokeAgentDetails(
    details=agent_details,        # AgentDetails instance
    endpoint=my_endpoint,         # Optional endpoint (with hostname/port)
    session_id="session-42"
)
tenant_details = TenantDetails(tenant_id="tenant-123")
req = Request(content="User asks a question")

with InvokeAgentScope.start(invoke_details, tenant_details, req):
    # Perform agent invocation logic
    response = call_agent(...)

import { 
  InvokeAgentScope, 
  ExecutionType, 
  CallerDetails, 
  EnhancedAgentDetails 
} from '@microsoft/agents-a365-observability';

// Basic agent invocation details
const invokeDetails = {
  agentId: 'email-agent-123',
  agentName: 'Email Assistant',
  ...
  }
};

const tenantDetails = {
  tenantId: 'tenant-789'
};

// Optional: Caller details (human user)
const callerDetails: CallerDetails = {};

// Optional: Caller agent details (for agent-to-agent calls)
const callerAgentDetails: EnhancedAgentDetails = {};

// Enhanced invocation with caller context
using scope = InvokeAgentScope.start(
  invokeDetails,
  tenantDetails,
  callerAgentDetails,
  callerDetails
);

try {
  // Record input messages
  scope.recordInputMessages(['Please help me organize my emails', 'Focus on urgent items']);
  
  // Your agent invocation logic here
  const response = await invokeAgent(invokeDetails.request.content);
  
  // Record output messages
  scope.recordOutputMessages(['I found 15 urgent emails', 'Here is your organized inbox']);
  
} catch (error) {
  scope.recordError(error as Error);
  throw error;
}
// Scope automatically disposed at end of using block

using System;
using System.Threading.Tasks;
using Microsoft.Agents.A365.Observability.Runtime.Tracing.Contracts;
using Microsoft.Agents.A365.Observability.Runtime.Tracing.Scopes;

public class MyAgent
{
    public async Task<AgentResponse> ProcessUserRequest(string userInput)
    {
        var agentDetails = new AgentDetails(
            agentId: Guid.NewGuid().ToString(),
            agentName: "MyAgent",
            agentDescription: "Handles user requests.",
            tenantId: "tenant-789"
        );

        var tenantDetails = new TenantDetails(Guid.Parse("11111111-2222-3333-4444-555555555555"));

        var request = new Request(
            content: userInput,
            executionType: ExecutionType.HumanToAgent,
            sessionId: "session-abc",
            channelMetadata: new ChannelMetadata("webchat", "https://webchat.contoso.com")
        );

        var callerDetails = new CallerDetails(
            callerId: "user-123",
            callerName: "Jane Doe",
            callerUpn: "jane.doe@contoso.com",
            callerUserId: "user-uuid-456",
            tenantId: "tenant-789"
        );

        var endpoint = new Uri("https://myagent.contoso.com");

        var invokeAgentDetails = new InvokeAgentDetails(endpoint, agentDetails, sessionId: "session-abc");

        var conversationId = "conv-xyz";

        // Start the scope
        using var scope = InvokeAgentScope.Start(
            invokeAgentDetails: invokeAgentDetails,
            tenantDetails: tenantDetails,
            request: request,
            callerAgentDetails: null,
            callerDetails: callerDetails,
            conversationId: conversationId
        );

        // Record input messages
        scope.RecordInputMessages(new[] { userInput });

        // ... your agent logic here ...

        // Simulate agent processing and output
        var output = $"Processed: {userInput}";
        scope.RecordOutputMessages(new[] { output });

        // Optionally record a single response
        scope.RecordResponse(output);

        return new AgentResponse { Content = output };
    }
}

public class AgentResponse
{
    public string Content { get; set; }
}

工具执行 ID

以下示例演示如何使用可观测性跟踪检测代理的工具执行，以捕获遥测数据以进行监视和审核。

from microsoft_agents_a365.observability.core.execute_tool_scope import ExecuteToolScope
from microsoft_agents_a365.observability.core.tool_call_details import ToolCallDetails

tool_details = ToolCallDetails(
    tool_name="summarize",
    tool_type="function",
    tool_call_id="tc-001",
    arguments="{'text': '...'}",
    description="Summarize provided text",
    endpoint=None  # or endpoint object with hostname/port
)

with ExecuteToolScope.start(tool_details, agent_details, tenant_details):
    result = run_tool(tool_details)

import { ExecuteToolScope } from '@microsoft/agents-a365-observability';

const toolDetails = {
  toolName: 'email-search',
  arguments: JSON.stringify({ query: 'from:boss@company.com', limit: 10 }),
  toolCallId: 'tool-call-456',
  description: 'Search emails by criteria',
  toolType: 'function',
  endpoint: {
    host: 'tools.contoso.com',
    port: 8080,  // Will be recorded since not 443
    protocol: 'https'
  }
};

using scope = ExecuteToolScope.start(toolDetails, agentDetails, tenantDetails);

try {
  // Execute the tool
  const result = await searchEmails(toolDetails.arguments);
  
  // Record the tool execution result
  scope.recordResponse(JSON.stringify(result));
  
  return result;
} catch (error) {
  scope.recordError(error as Error);
  throw error;
}

using System;
using System.Threading.Tasks;
using Microsoft.Agents.A365.Observability.Runtime.Tracing.Contracts;
using Microsoft.Agents.A365.Observability.Runtime.Tracing.Scopes;

public class MyToolAgent
{
    public async Task<ToolResult> ExecuteTool(string toolName, object parameters)
    {
        var agentDetails = new AgentDetails(
            agentId: Guid.NewGuid().ToString(),
            agentName: "ToolAgent",
            agentDescription: "Executes tools for users.",
            agentAUID: "tool-auid-123",
            agentUPN: "toolagent@contoso.com",
            agentBlueprintId: "tool-blueprint-456",
            agentType: AgentType.EntraEmbodied,
            tenantId: "tenant-789"
        );

        var tenantDetails = new TenantDetails(Guid.Parse("11111111-2222-3333-4444-555555555555"));

        var endpoint = new Uri("https://toolagent.contoso.com:8443");

        var toolCallDetails = new ToolCallDetails(
            toolName: toolName,
            arguments: System.Text.Json.JsonSerializer.Serialize(parameters),
            toolCallId: Guid.NewGuid().ToString(),
            description: "Runs a tool operation.",
            toolType: "custom-type",
            endpoint: endpoint
        );

        // Start the scope
        using var scope = ExecuteToolScope.Start(
            toolCallDetails: toolCallDetails,
            agentDetails: agentDetails,
            tenantDetails: tenantDetails
        );

        // ... your tool logic here ...

        // Record response
        var toolOutput = $"Tool '{toolName}' processed with parameters: {System.Text.Json.JsonSerializer.Serialize(parameters)}";
        scope.RecordResponse(toolOutput);

        return new ToolResult { Output = toolOutput };
    }
}

public class ToolResult
{
    public string Output { get; set; }
}

推理

以下示例演示如何使用可观测性跟踪检测 AI 模型推理调用，以捕获令牌使用情况、模型详细信息和响应元数据。

from microsoft_agents_a365.observability.core.inference_scope import InferenceScope
from microsoft_agents_a365.observability.core.inference_call_details import InferenceCallDetails
from microsoft_agents_a365.observability.core.request import Request

inference_details = InferenceCallDetails(
    operationName=SomeEnumOrValue("chat"),
    model="gpt-4o-mini",
    providerName="azure-openai",
    inputTokens=123,
    outputTokens=456,
    finishReasons=["stop"],
    responseId="resp-987"
)
req = Request(content="Explain quantum computing simply.")

with InferenceScope.start(inference_details, agent_details, tenant_details, req):
    completion = call_llm(...)

import { InferenceScope, InferenceOperationType } from '@microsoft/agents-a365-observability';

const inferenceDetails = {
  operationName: InferenceOperationType.CHAT,
  model: 'gpt-4',
  providerName: 'openai',
  inputTokens: 150,
  outputTokens: 75,
  finishReasons: ['stop'],
  responseId: 'resp-123456'
};

using scope = InferenceScope.start(inferenceDetails, agentDetails, tenantDetails);

try {
  // Record input messages
  scope.recordInputMessages(['Summarize the following emails for me...']);
  
  // Call the LLM
  const response = await callLLM();
  
  // Record detailed telemetry with granular methods
  scope.recordOutputMessages(['Here is your email summary...']);
  scope.recordInputTokens(145);  // Update if different from constructor
  scope.recordOutputTokens(82);  // Update if different from constructor
  scope.recordResponseId('resp-789123');
  scope.recordFinishReasons(['stop', 'max_tokens']);
  
  return response.text;
} catch (error) {
  scope.recordError(error as Error);
  throw error;
}

using System;
using System.Threading.Tasks;
using Microsoft.Agents.A365.Observability.Runtime.Tracing.Contracts;
using Microsoft.Agents.A365.Observability.Runtime.Tracing.Scopes;

public class MyInferenceAgent
{
    public async Task<InferenceResult> RunInference(string input)
    {
        var agentDetails = new AgentDetails(
            agentId: Guid.NewGuid().ToString(),
            agentName: "InferenceAgent",
            agentDescription: "Performs generative AI inference.",
            agentAUID: "inference-auid-123",
            agentUPN: "inferenceagent@contoso.com",
            agentBlueprintId: "inference-blueprint-456",
            agentType: AgentType.EntraEmbodied,
            tenantId: "tenant-789"
        );

        var tenantDetails = new TenantDetails(Guid.Parse("11111111-2222-3333-4444-555555555555"));

        var inferenceDetails = new InferenceCallDetails(
            operationName: InferenceOperationType.Chat,
            model: "gpt-4",
            providerName: "OpenAI",
            inputTokens: 42,
            outputTokens: 84,
            finishReasons: new[] { "stop", "length" },
            responseId: "response-xyz"
        );

        // Start the scope
        using var scope = InferenceScope.Start(
            details: inferenceDetails,
            agentDetails: agentDetails,
            tenantDetails: tenantDetails
        );

        // ... your inference logic here ...

        // Record input/output messages and other telemetry
        scope.RecordInputMessages(new[] { input, "additional context" });
        scope.RecordOutputMessages(new[] { "AI response message" });
        scope.RecordInputTokens(42);
        scope.RecordOutputTokens(84);
        scope.RecordResponseId("response-xyz");
        scope.RecordFinishReasons(new[] { "stop", "length" });
        scope.RecordThoughtProcess("Reasoning step 1; step 2");

        return new InferenceResult { Output = "AI response message" };
    }
}

public class InferenceResult
{
    public string Output { get; set; }
}

在本地验证

将 ENABLE_A365_OBSERVABILITY_EXPORTER 环境变量设置为 false。此导出范围（跟踪）到控制台。

使用可观测性测试代理

在代理中实现可观测性后，测试以确保正确捕获遥测数据。按照测试指南设置环境，然后主要关注“查看可观测性日志”部分，以验证可观测性实现是否按预期工作。

反馈

此页面是否有帮助？

Last updated on 2025-12-02

通过

可观测性代理

关键优势

安装

配置

行李属性

令牌解析程序

自动检测

语义内核

OpenAI

代理框架

LangChain 框架

手动检测

代理调用

工具执行 ID

推理

在本地验证

使用可观测性测试代理

反馈

其他资源