LangChain Agent MCP Server

# Examples and Tutorials ## Quick Examples ### Example 1: Build a RAG Agent in 10 Lines Here's how to create a simple RAG (Retrieval-Augmented Generation) agent using the MCP server: ```python # Simple RAG agent example from langchain_openai import ChatOpenAI from langchain.tools import Tool from langchain.agents import AgentExecutor, create_react_agent from langchain import hub # 1. Initialize LLM llm = ChatOpenAI(model="gpt-4o-mini", temperature=0) # 2. Define a simple search tool (replace with your RAG retrieval) def search_docs(query: str) -> str: # Your RAG retrieval logic here return f"Retrieved: {query}" # 3. Create tool tools = [Tool(name="search", func=search_docs, description="Search documents")] # 4. Get prompt and create agent prompt = hub.pull("hwchase17/react") agent = create_react_agent(llm, tools, prompt) # 5. Create executor rag_agent = AgentExecutor(agent=agent, tools=tools, verbose=True) # 6. Use it! result = rag_agent.invoke({"input": "What is in the documents about AI?"}) print(result["output"]) ``` **That's it!** 10 lines of core code to build a RAG agent. ### Example 2: Custom Tool Integration Add your own tools to the agent: ```python # In src/agent.py def my_custom_tool(query: str) -> str: """Your custom tool description.""" # Your implementation return result def create_agent_tools() -> List[Tool]: tools = [ Tool( name="my_custom_tool", func=my_custom_tool, description="Description for the LLM" ), # ... existing tools ] return tools ``` ### Example 3: Using the MCP Server Once deployed, use the server from any MCP client: ```python import requests # Service URL service_url = "https://langchain-agent-mcp-server-554655392699.us-central1.run.app" # Invoke the agent response = requests.post( f"{service_url}/mcp/invoke", json={ "tool": "agent_executor", "arguments": { "query": "What is the weather in New York?" } } ) result = response.json() print(result["content"][0]["text"]) ``` ### Example 4: SlashMCP Integration Register and use in SlashMCP: ``` /slashmcp add langchain-agent https://langchain-agent-mcp-server-554655392699.us-central1.run.app /srv_xxxxx agent_executor query="Your question here" ``` ## Advanced Examples ### Building a Document Q&A Agent ```python from langchain_openai import ChatOpenAI from langchain.tools import Tool from langchain.agents import AgentExecutor, create_react_agent from langchain import hub llm = ChatOpenAI(model="gpt-4o-mini") def search_documents(query: str) -> str: """Search through your document database.""" # Implement your vector search or database query # This is a placeholder return f"Found documents related to: {query}" def extract_answer(context: str, question: str) -> str: """Extract answer from context.""" # Use LLM to extract answer return f"Answer based on: {context}" tools = [ Tool(name="search_documents", func=search_documents, description="Search document database"), Tool(name="extract_answer", func=extract_answer, description="Extract answer from context") ] prompt = hub.pull("hwchase17/react") agent = create_react_agent(llm, tools, prompt) executor = AgentExecutor(agent=agent, tools=tools) result = executor.invoke({"input": "What did the report say about revenue?"}) ``` ### Multi-Step Reasoning Agent ```python # Agent that performs multiple steps def get_data(query: str) -> str: """Fetch data from API.""" return "Data: 1000 units" def analyze_data(data: str) -> str: """Analyze the data.""" return f"Analysis: {data} shows growth" def generate_report(analysis: str) -> str: """Generate final report.""" return f"Report: {analysis}" tools = [ Tool(name="get_data", func=get_data, description="Fetch data"), Tool(name="analyze_data", func=analyze_data, description="Analyze data"), Tool(name="generate_report", func=generate_report, description="Generate report") ] # Agent will chain these tools together automatically ``` ## Integration Examples ### FastAPI Client ```python from fastapi import FastAPI import httpx app = FastAPI() @app.post("/ask") async def ask_question(question: str): async with httpx.AsyncClient() as client: response = await client.post( "https://langchain-agent-mcp-server-554655392699.us-central1.run.app/mcp/invoke", json={ "tool": "agent_executor", "arguments": {"query": question} } ) return response.json() ``` ### Python Script ```python import requests def ask_agent(question: str): url = "https://langchain-agent-mcp-server-554655392699.us-central1.run.app/mcp/invoke" response = requests.post( url, json={ "tool": "agent_executor", "arguments": {"query": question} } ) return response.json()["content"][0]["text"] # Use it answer = ask_agent("What is machine learning?") print(answer) ``` --- **Want more examples?** Check out the [API Reference](api-reference.md) for detailed endpoint documentation.