MCP Server

🚀 LangChain MCP Multi-Server Integration

A comprehensive Model Context Protocol (MCP) implementation that integrates multiple data sources and services with LangChain agents for intelligent data processing and storage.

🚀 Quick Start

This project offers a comprehensive MCP implementation, integrating various data - sources and services with LangChain agents. It enables intelligent data processing and storage.

✨ Features

• Multi - Server Architecture: Incorporates Weather, Stocks, MongoDB, and Elasticsearch servers.
• AI Agent Integration: Employs a LangChain agent that can intelligently utilize all available tools.
• Automatic Data Storage: Automatically stores conversations and data in Elasticsearch.
• Real - time Data: Provides live weather alerts and stock portfolio data.
• Secure Configuration: Manages credentials via .netrc files.

📦 Installation

1. Clone the repository:

git clone https://github.com/vamsivenkat987/langchain_mcp_server.git
cd langchain_mcp_server/mcp

2. Install dependencies:

pip install langchain - mcp - adapters langchain - openai elasticsearch pymongo kiteconnect requests

3. Start required services:

# Start Elasticsearch
docker run -d -p 9200:9200 -e "discovery.type=single - node" elasticsearch:8.11.0

# Start MongoDB (optional)
docker run -d -p 27017:27017 mongo:latest

⚙️ Configuration

Create a .netrc file in the project directory:

machine llm.model
login your_api_key_here
password https://your - llm - endpoint - here

machine stocks.data
login your_kite_api_key
password your_kite_api_secret
account your_kite_access_token

💻 Usage Examples

Basic Usage

python3 mcp_client.py

Advanced Usage

# Test weather server
python3 weather_server.py

# Test stock server
python3 stocks_server.py

# Test Elasticsearch server
python3 elastic_search_server.py

# Test MongoDB server
python3 mongo_server.py

🔧 Technical Details

Available Tools

Weather Server (weather_server.py)

• get_alerts(state) - Retrieve active weather alerts for a US state. It uses the National Weather Service API.

Stock Server (stocks_server.py)

• get_holdings() - Fetch the current stock portfolio from Kite Connect. A valid Kite Connect session is required.

Elasticsearch Server (elastic_search_server.py)

• add_document(document, index, doc_id) - Store documents in Elasticsearch. It automatically generates timestamps and UUIDs and parses JSON for string inputs.

MongoDB Server (mongo_server.py)

• add_conversation_data(user_content, ai_content) - Store conversations.
• add_customer_data(name, age) - Store customer information.

Project Structure

mcp/
├── mcp_client.py              # Main LangChain agent client
├── weather_server.py          # Weather alerts MCP server
├── stocks_server.py           # Stock portfolio MCP server
├── elastic_search_server.py   # Elasticsearch MCP server
├── mongo_server.py            # MongoDB MCP server
├── test_elastic_search.py     # Elasticsearch testing utilities
├── test_kite.py              # Kite Connect authentication flow
├── test_mcp.py               # MCP server testing
├── .netrc                    # API credentials (create this)
├── oneEGaDCA.crt            # SSL certificate
└── README.md                 # This file

AI Agent Workflow

The LangChain agent follows this workflow:

1. Process User Query: Comprehend the user's question.
2. Call Appropriate Tools: For weather queries, call get_alerts(state); for stock queries, call get_holdings().
3. Provide Natural Response: Answer the user's question in plain language.
4. Store Data: Automatically save conversations and data to Elasticsearch.

Data Storage

Elasticsearch Indices

• conversations - User queries and AI responses.
• weather - Weather alert data.
• stocks - Stock portfolio information.
• customer_data - Customer information.

MongoDB Collections

• conversations - Chat history.
• customers - Customer data.

Security Features

• Credentials are stored in .netrc files, not in the code.
• Supports SSL certificates.
• Allows environment variable configuration.
• Handles API tokens securely.

📚 Documentation

Example Queries

• "Get me the current weather alerts for New York"
• "What are my stock holdings?"
• "Is there any connection between my stock portfolio and weather in NY?"
• "Show me the weather in California and save the data"

Troubleshooting

Common Issues

1. Elasticsearch Connection Error: Ensure Elasticsearch is running on localhost:9200 and check its accessibility with curl http://localhost:9200.
2. Kite Connect Authentication: Verify API credentials in .netrc and check if the access token is valid (tokens expire daily).
3. MCP Server Errors: Check server logs for detailed error messages and ensure all dependencies are installed.
4. LLM Timeout Errors: Increase the timeout in mcp_client.py or try using streaming instead of ainvoke.

API Documentation

Weather Server

• Endpoint: National Weather Service API
• Rate Limits: No authentication required
• Data Format: JSON weather alerts

Stock Server

• Endpoint: Kite Connect API
• Authentication: API key + secret + access token
• Rate Limits: As per Zerodha limits
• Data Format: Portfolio holdings with P&L

🤝 Contributing

1. Fork the repository.
2. Create a feature branch.
3. Make your changes.
4. Add tests if applicable.
5. Submit a pull request.

🆘 Support

For issues and questions:

• Create an issue on GitHub.
• Check the troubleshooting section.
• Review server logs for detailed error information.

🔄 Version History

• v1.0.0 - Initial release with multi - server MCP integration.
• v1.1.0 - Added automatic data storage and improved error handling.
• v1.2.0 - Enhanced authentication and security features.