🤖 LangChain TypeScript Examples

This repository demonstrates different ways to integrate LangChain with TypeScript, including native tool calling, single MCP server, and multiple MCP servers.

🚀 Getting Started

1. Clone the Repository

git clone https://github.com/rehmat123/Langchain-typescript
cd langchain-typescript

2. Install Dependencies

pnpm install

3. Configure Environment

Create a .env file in the root:

OPENAI_API_KEY=your-openai-key

4. Configure Cursor with Weather MCP Server

edit the MCP configuration file at ~/.cursor/mcp.json:

{
  "mcpServers": {
    "weather": {
      "command": "npx",
      "args": ["-y", "@rehmatalisayany/weather-mcp-server"],
      "transport": "stdio"
    },
  }
}

📚 LangChain Integrations

1. Native Tool Calling

The src/langchain/toolsCalling.ts file demonstrates how to use LangChain with native tool calling. This approach allows you to define custom tools directly in your code.

pnpm function-calling-langchain

2. Single MCP Server

The src/langchain/singleMcp.ts file shows how to integrate a single MCP server (weather service) with LangChain. This allows you to use external services as tools in your LangChain agents.

pnpm single-mcp-weather-langchain

3. Multiple MCP Servers

The src/langchain/multiMcp.ts file demonstrates how to use multiple MCP servers (weather and math) with LangChain. This enables your agents to use different tools from various services.

pnpm run multi-mcp-weather-langchain

🧰 Tech Stack

• LangChain.js - Framework for building context-aware, language model-driven applications
• TypeScript - Strongly typed JavaScript for better developer experience
• OpenAI - Used as the underlying language model
• MCP Server - Model Context Protocol server for external services

Project Structure

.
├── src/
│   ├── langchain/
│   │   ├── toolsCalling.ts  # Native tool calling example
│   │   ├── singleMcp.ts     # Single MCP server example
│   │   └── multiMcp.ts      # Multiple MCP servers example
│   └── index.ts             # Main entry point
├── package.json
└── README.md

Common Issues

1. MCP Server Connection Timeout

   • If you see errors like Failed to connect to stdio server: McpError: MCP error -32001: Request timed out, try:
     • Reducing the delayMs value in the server configuration
     • Ensuring the MCP server package is installed correctly
     • Checking your network connection

2. Tool Not Found

   • If you see errors about tools not being found, ensure:
     • The tool name in your prompt matches the actual tool name (including prefixes)
     • The MCP server is properly configured and running

3. Template Variable Errors

   • If you see errors like Missing value for input variable, check:
     • Your prompt template for proper variable formatting
     • That all required variables are provided in the invoke method

Debugging Tips

• Enable verbose output in the agent executor to see detailed logs
• Check the console for specific error messages
• Verify that all required environment variables are set

🔧 Configuration

The project uses the following configuration for MCP servers:

{
  "mcpServers": {
    "weather": {
      "command": "npx",
      "args": ["-y", "@rehmatalisayany/weather-mcp-server"],
      "transport": "stdio"
    },
    "math": {
      "command": "npx",
      "args": ["-y", "nm-mcp-math"],
      "transport": "stdio"
    }
  }
}

🐛 Troubleshooting

If you encounter issues:

1. Make sure your OpenAI API key is correctly set in the .env file
2. Verify that all dependencies are installed
3. Check that the MCP servers are running correctly
4. For connection timeouts, try reducing the delayMs value in the server configuration

License

MIT