> For the complete documentation index, see [llms.txt](https://docs.talordata.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.talordata.com/cn-tw/serp-api/integration/mcp-integration/langchain-ji-cheng.md).

# LangChain集成

将 LangChain 和 LangGraph 代理连接到 TalorData MCP 服务器，即可将实时网络搜索、抓取和结构化数据工具添加到您的 AI 工作流程中。

### 托管 MCP

{% stepper %}
{% step %}
**获取您的 API 令牌**

* 前往 [**TalorData SERP API - API Token**](https://dashboard.talordata.com/scraping/serp-api/api-token) 创建API Token；
* 复制您的 API 令牌（格式如下：`sk_7qBRe***************************`）
  {% endstep %}

{% step %}
**安装所需软件包**

```
pip install langchain-mcp-adapters
```

{% endstep %}

{% step %}
**配置您的 MCP 服务器**

```
import asyncio
from langchain_openai import ChatOpenAI
from langchain.agents import create_agent
from langchain_mcp_adapters.client import MultiServerMCPClient
from dotenv import load_dotenv
import os

load_dotenv()

async def main():
    # Configure MCP client
    client = MultiServerMCPClient({
        "talor_data": {
            "url": "https://mcp.talordata.net/<API_TOKEN>/mcp",
            "transport": "http",
        }
    })

    # Get available tools
    tools = await client.get_tools()
    print("Available tools:", [tool.name for tool in tools])

    # Configure LLM
    llm = ChatOpenAI(
        openai_api_key=os.getenv("OPENROUTER_API_KEY"),
        openai_api_base="https://openrouter.ai/api/v1",
        model_name="moonshotai/kimi-k2"
    )

    # System prompt for web search agent
    system_prompt = """
    You are a web search agent with comprehensive scraping capabilities. Your tools include:
    - **search_engine**: Get search results from Google/Bing/Yandex
    - **scrape_as_markdown**: Extract content from any webpage with bot detection bypass
    - **Structured extractors**: Fast, reliable data from major platforms (Amazon, LinkedIn, Instagram, Facebook, X, TikTok, YouTube, Reddit, Zillow, etc.)
    - **Browser automation**: Navigate, click, type, screenshot for complex interactions

    Guidelines:
    - Use structured web_data_* tools for supported platforms when possible (faster/more reliable)
    - Use general scraping for other sites
    - Handle errors gracefully and respect rate limits
    - Think step by step about what information you need and which tools to use
    - Be thorough in your research and provide comprehensive answers

    When responding, follow this pattern:
    1. Think about what information is needed
    2. Choose the appropriate tool(s)
    3. Execute the tool(s)
    4. Analyze the results
    5. Provide a clear, comprehensive answer
    """

    # Create ReAct agent
    agent = create_agent(
        model=llm,
        tools=tools,
        system_prompt=system_prompt
    )

    # Test the agent
    print("Testing ReAct Agent with available tools...")
    print("=" * 50)

    result = await agent.ainvoke({
        "messages": [("human", "Search for the latest news about AI developments")]
    })

    print("\nAgent Response:")
    print(result["messages"][-1].content)

if __name__ == "__main__":
    asyncio.run(main())
```

{% endstep %}

{% step %}
**设置环境变量**

`.env`在项目目录中创建一个文件：

```
OPENROUTER_API_KEY=your_openrouter_api_key_here
```

{% endstep %}

{% step %}
**测试一下是否有效**

* 请替换`<API_TOKEN>`为您的实际 TalorData API 令牌
* 运行你的 LangChain 脚本
* 你应该看到代理人执行网络搜索并提供全面的回复。
  {% endstep %}

{% step %}
**监控使用情况**

* 在 TalorData SERP API-[**使用记录中**](https://dashboard.talordata.com/scraping/serp-api/statistics)查看您的 API 使用情况。
  {% endstep %}
  {% endstepper %}

如需更多協助，請透過 **線上客服** 或電郵 **<support@talordata.com>** 聯絡我們。