How to Integrate Oxylabs Web Scraper API With LangGraph

brew install uv

#!/usr/bin/env python3


import os
import asyncio
from dotenv import load_dotenv
from langchain_mcp_adapters.sessions import create_session
from langchain_mcp_adapters.tools import load_mcp_tools
from langgraph_supervisor import create_supervisor
from langgraph.prebuilt import create_react_agent
from langchain_openai import ChatOpenAI




load_dotenv()


# Define the MCP server config
config = {
   "transport": "stdio",
   "command": "uvx",
   "args": ["oxylabs-mcp"],
   "env": {
       "OXYLABS_USERNAME": os.getenv("OXYLABS_USERNAME"),
       "OXYLABS_PASSWORD": os.getenv("OXYLABS_PASSWORD"),
   }
}


async def main():
   # Initialize an MCP server session and load the tools
   async with create_session(config) as session:
       await session.initialize()
       tools = await load_mcp_tools(session)


       # Create an AI agent that uses the MCP server tools
   research_agent = create_react_agent(
       model="openai:gpt-4.1",
       tools=tools,
       name="research_agent",
       prompt="You are a research agent helping another LLM agent to find information about a given query. You have access to a search tool that can search the web for information. You will use the search tool to find information about the query. You will then use the information to help the other LLM agent. You will only use the search tool to find information about the query. You will not use the search tool to find information about anything else."
   )


   # Initialize the model for supervisor
   supervisor_model = ChatOpenAI(model="gpt-4.1", temperature=0)


   # Create supervisor workflow
   workflow = create_supervisor(
       agents=[research_agent],
       model=supervisor_model,
       prompt=(
           "You are a Company Intelligence Supervisor responsible for orchestrating comprehensive company data enrichment processes. "
           "Your primary objective is to create detailed company profiles by coordinating with your research agent.\n\n"
          
           "CORE MISSION:\n"
           "Transform basic company information into comprehensive business intelligence profiles through systematic data enrichment.\n\n"
          
           "WORKFLOW PROCESS:\n"
           "1. INITIAL ANALYSIS: Parse the company input (name, domain, or basic info)\n"
           "2. RESEARCH COORDINATION: Direct your research agent to gather comprehensive company data\n"
           "3. DATA SYNTHESIS: Organize findings into a structured company profile\n"
           "4. QUALITY ASSURANCE: Ensure completeness and accuracy of the enriched profile\n"
           "5. FINAL DELIVERY: Present a professional company intelligence report\n\n"
          
           "RESEARCH AREAS TO COVER:\n"
           "• Company Overview: Legal name, founding date, headquarters, business model\n"
           "• Financial Information: Revenue, funding rounds, valuation, financial health\n"
           "• Industry & Market: Sector classification, market position, competitors\n"
           "• Leadership: Key executives, founders, board members\n"
           "• Products & Services: Core offerings, product lines, key differentiators\n"
           "• Technology Stack: Known technologies, platforms, technical infrastructure\n"
           "• Recent News: Latest developments, partnerships, acquisitions, press releases\n"
           "• Contact Information: Official website, social media, contact details\n"
           "• Employee Information: Company size, key departments, hiring trends\n"
           "• Regulatory & Compliance: Industry regulations, certifications, legal status\n\n"
          
           "DELEGATION INSTRUCTIONS:\n"
           "Always delegate research tasks to your research_agent with specific, targeted queries. "
           "Break down complex research into multiple focused searches for comprehensive coverage.\n\n"
          
           "OUTPUT FORMAT:\n"
           "Deliver results as a structured company profile with clear sections, bullet points, and actionable intelligence. "
           "Include data sources and confidence levels where appropriate.\n\n"
          
           "QUALITY STANDARDS:\n"
           "• Accuracy: Verify information through multiple sources when possible\n"
           "• Completeness: Address all requested profile areas\n"
           "• Timeliness: Focus on recent and current information\n"
           "• Relevance: Prioritize actionable business intelligence\n"
           "• Professional Format: Present findings in a clear, executive-ready format"
       )
   )


   app = workflow.compile()


       # A loop to ask questions and get answers
   while True:
       question = input("\nQuestion -> ")
       if question == "exit":
           break
       result = await app.ainvoke({"messages": question})
       print(f"\n{result['messages'][-1].content}\n")


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