LangSmith 可以通过 OpenInference 的 AutoGen 插桩功能捕获由 AutoGen 生成的追踪数据。本指南将向您展示如何自动捕获来自 AutoGen 多智能体对话的追踪数据，并将其发送到 LangSmith 进行监控和分析。

安装

使用您偏好的包管理器安装所需的包：

pip

pip install langsmith autogen openinference-instrumentation-autogen openinference-instrumentation-openai

uv

uv add langsmith autogen openinference-instrumentation-autogen openinference-instrumentation-openai

为了获得最佳的 OpenTelemetry 支持，需要 LangSmith Python SDK 版本 langsmith>=0.4.26。

设置

1. 配置环境变量

设置您的 API 密钥和项目名称：

Shell

export LANGSMITH_API_KEY=<your_langsmith_api_key>
export LANGSMITH_PROJECT=<your_project_name>
export OPENAI_API_KEY=<your_openai_api_key>

2. 配置 OpenTelemetry 集成

在您的 AutoGen 应用程序中，导入并配置 LangSmith 的 OpenTelemetry 集成，以及 AutoGen 和 OpenAI 的插桩器：

from langsmith.integrations.otel import configure
from openinference.instrumentation.autogen import AutogenInstrumentor
from openinference.instrumentation.openai import OpenAIInstrumentor

# 配置 LangSmith 追踪
configure(project_name="autogen-demo")

# 插桩 AutoGen 和 OpenAI 调用
AutogenInstrumentor().instrument()
OpenAIInstrumentor().instrument()

您无需手动设置任何 OpenTelemetry 环境变量或配置导出器——configure() 会自动处理所有事情。

3. 创建并运行您的 AutoGen 应用程序

配置完成后，您的 AutoGen 应用程序将自动向 LangSmith 发送追踪数据：

import autogen
from openinference.instrumentation.autogen import AutogenInstrumentor
from openinference.instrumentation.openai import OpenAIInstrumentor
from langsmith.integrations.otel import configure
import os
import dotenv

# 加载环境变量
dotenv.load_dotenv(".env.local")

# 配置 LangSmith 追踪
configure(project_name="autogen-code-review")

# 插桩 AutoGen 和 OpenAI
AutogenInstrumentor().instrument()
OpenAIInstrumentor().instrument()

# 配置您的智能体
config_list = [
    {
        "model": "gpt-4",
        "api_key": os.getenv("OPENAI_API_KEY"),
    }
]

# 创建代码审查员智能体
code_reviewer = autogen.AssistantAgent(
    name="code_reviewer",
    llm_config={"config_list": config_list},
    system_message="""You are an expert code reviewer. Your role is to:
    1. Review code for bugs, security issues, and best practices
    2. Suggest improvements and optimizations
    3. Provide constructive feedback
    Always be thorough but constructive in your reviews.""",
)

# 创建开发者智能体
developer = autogen.AssistantAgent(
    name="developer",
    llm_config={"config_list": config_list},
    system_message="""You are a senior software developer. Your role is to:
    1. Write clean, efficient code
    2. Address feedback from code reviews
    3. Explain your implementation decisions
    4. Implement requested features and fixes""",
)

# 创建用户代理智能体
user_proxy = autogen.UserProxyAgent(
    name="user_proxy",
    human_input_mode="NEVER",
    max_consecutive_auto_reply=8,
    is_termination_msg=lambda x: x.get("content", "").rstrip().endswith("TERMINATE"),
    code_execution_config={"work_dir": "workspace"},
    llm_config={"config_list": config_list},
)

def run_code_review_session(task_description: str):
    """Run a multi-agent code review session."""

    # 使用智能体创建群聊
    groupchat = autogen.GroupChat(
        agents=[user_proxy, developer, code_reviewer],
        messages=[],
        max_round=10
    )

    # 创建群聊管理器
    manager = autogen.GroupChatManager(
        groupchat=groupchat,
        llm_config={"config_list": config_list}
    )

    # 开始对话
    user_proxy.initiate_chat(
        manager,
        message=f"""
        Task: {task_description}

        Developer: Please implement the requested feature.
        Code Reviewer: Please review the implementation and provide feedback.

        Work together to create a high-quality solution.
        """
    )

    return "Code review session completed"

# 示例用法
if __name__ == "__main__":
    task = """
    Create a Python function that implements a binary search algorithm.
    The function should:
    - Take a sorted list and a target value as parameters
    - Return the index of the target if found, or -1 if not found
    - Include proper error handling and documentation
    """

    result = run_code_review_session(task)
    print(f"Result: {result}")

高级用法

自定义元数据和标签

您可以通过在 AutoGen 应用程序中设置跨度（span）属性来为追踪数据添加自定义元数据：

from opentelemetry import trace

# 获取当前的追踪器
tracer = trace.get_tracer(__name__)

def run_code_review_session(task_description: str):
    with tracer.start_as_current_span("autogen_code_review") as span:
        # 添加自定义元数据
        span.set_attribute("langsmith.metadata.session_type", "code_review")
        span.set_attribute("langsmith.metadata.agent_count", "3")
        span.set_attribute("langsmith.metadata.task_complexity", "medium")
        span.set_attribute("langsmith.span.tags", "autogen,code-review,multi-agent")

        # 您的 AutoGen 代码放在这里
        groupchat = autogen.GroupChat(
            agents=[user_proxy, developer, code_reviewer],
            messages=[],
            max_round=10
        )

        manager = autogen.GroupChatManager(
            groupchat=groupchat,
            llm_config={"config_list": config_list}
        )

        user_proxy.initiate_chat(manager, message=task_description)
        return "Session completed"

与其他插桩器结合使用

您可以将 AutoGen 插桩与其他插桩器（例如 Semantic Kernel、DSPy）结合使用，只需将它们添加并初始化为插桩器即可：

from langsmith.integrations.otel import configure
from openinference.instrumentation.autogen import AutogenInstrumentor
from openinference.instrumentation.openai import OpenAIInstrumentor
from openinference.instrumentation.dspy import DSPyInstrumentor

# 配置 LangSmith 追踪
configure(project_name="multi-framework-app")

# 初始化多个插桩器
AutogenInstrumentor().instrument()
OpenAIInstrumentor().instrument()
DSPyInstrumentor().instrument()

# 您使用多个框架的应用程序代码