LangChain Agentic Platform with MCP Integration

A modular agentic platform with multi-agent coordination and Model Context Protocol (MCP) integration for connecting AI agents to external tools and services.

🚀 Quick Start

Prerequisites

• Node.js 18+
• Python 3.8+
• pnpm (recommended) or npm

Installation

1. Clone and install dependencies:

git clone <repository-url>
cd langchain-agent
pnpm install

2. Install MCP server dependencies:

cd mcp_servers
pip install -r requirements.txt
cd ..

3. Set up environment variables:

cp .env.example .env
# Edit .env with your API keys and configuration

Running the Platform

Option 1: Full Agentic Platform (Recommended)

# Build the project
pnpm run build

# Start the platform with MCP integration
pnpm start

This will start the full AgenticPlatform with:

• Coordinator agent for intelligent routing
• Multiple specialized agents (Oracle, Analyst, Degen)
• MCP plugins for external tools
• Terminal client for interaction

Option 2: Development Mode

# Run in development mode with hot reload
pnpm run dev

🔧 MCP Server Management

Available MCP Servers

The platform includes several pre-configured MCP servers:

1. Math Server

Location: mcp_servers/math_server.py Tools: Basic arithmetic, power, square root, expression evaluation

Manual Testing:

cd mcp_servers
python math_server.py
# Server runs on stdio transport (no manual interaction needed)

2. Task Management Server

Location: mcp_servers/task_management_server.py Tools: Create tasks, list tasks, get statistics, generate suggestions

Manual Testing:

cd mcp_servers
python task_management_server.py
# Server runs on stdio transport

3. Weather Server

Location: mcp_servers/weather_server.py Tools: Get weather, forecasts, temperature, alerts

Manual Testing:

cd mcp_servers
python weather_server.py
# Server runs on SSE transport (requires web server)

Starting MCP Servers Manually

For stdio transport (Math & Task Management):

cd mcp_servers
python math_server.py &
python task_management_server.py &

For SSE transport (Weather):

cd mcp_servers
python weather_server.py &
# Server will be available at http://localhost:8000/sse

Testing MCP Servers

You can test the MCP integration directly:

# Test MCP plugin functionality
pnpm run build
node dist/examples/test_mcp_plugin.js

🤖 Using the Agentic Platform

Starting the Platform

# Build and start
pnpm run build && pnpm start

Available Commands

Once the platform is running, you can interact with it through the terminal client:

General Knowledge (Oracle Agent)

• "What's the weather in San Francisco?"
• "Search for information about AI trends"
• "Calculate 15 * 23" (via Math MCP)
• "What is the square root of 144?" (via Math MCP)

Analysis (Analyst Agent)

• "Analyze this startup: [description]"
• "Evaluate the market potential of this project"
• "Create a task: Build a web app" (via Task MCP)
• "Generate task suggestions for mobile app" (via Task MCP)

Crypto/Blockchain (Degen Agent)

• "Should I buy this token?"
• "How do I make an onchain transaction?"

Task Management (via MCP)

• "Create a task: Implement user authentication, high priority"
• "Get all tasks"
• "Show project statistics"
• "Complete task 1"

Math Operations (via MCP)

• "Calculate 3 + 5"
• "What is 10 to the power of 3?"
• "Calculate the square root of 256"
• "Evaluate expression: (15 + 7) * 3"

Agent Routing

The platform automatically routes your requests to the most appropriate agent:

• Oracle: General knowledge, searches, math operations
• Analyst: Financial analysis, startup evaluation, task management
• Degen: Crypto trading, onchain transactions

📁 Project Structure

langchain-agent/
├── src/
│   ├── core/                 # Core agent and workflow logic
│   ├── plugins/              # Plugin system including MCP
│   ├── clients/              # Client interfaces
│   ├── config/               # Configuration files
│   └── examples/             # Example implementations
├── mcp_servers/              # MCP server implementations
│   ├── math_server.py        # Math operations server
│   ├── task_management_server.py  # Task management server
│   ├── weather_server.py     # Weather information server
│   └── requirements.txt      # Python dependencies
├── docs/                     # Documentation
└── dist/                     # Compiled JavaScript

🔌 MCP Plugin Configuration

Using Factory Methods

import { MCPPluginFactory } from './plugins/MCPPlugin.js';

// Quick setup for common servers
const mathMCP = MCPPluginFactory.createMathPlugin();
const taskMCP = MCPPluginFactory.createTaskManagementPlugin();
const weatherMCP = MCPPluginFactory.createWeatherPlugin();

Custom Configuration

import { MCPPlugin } from './plugins/MCPPlugin.js';

const customMCP = new MCPPlugin({
  name: 'my_mcp',
  description: 'Custom MCP plugin',
  servers: [
    {
      name: 'my_server',
      command: 'python',
      args: ['./mcp_servers/my_server.py'],
      transport: 'stdio'
    }
  ]
});

🛠️ Development

Building the Project

# Build TypeScript to JavaScript
pnpm run build

# Watch mode for development
pnpm run build -- --watch

Running Examples

# MCP plugin examples
node dist/examples/mcp_plugin_example.js

# Test MCP functionality
node dist/examples/test_mcp_plugin.js

# Coordinator agent example
node dist/examples/coordinator_agent.js

Creating Custom MCP Servers

1. Create a new server file:

#!/usr/bin/env python3
from mcp.server.fastmcp import FastMCP

mcp = FastMCP("MyServer")

@mcp.tool()
def my_tool(input: str) -> str:
    """My custom tool."""
    return f"Processed: {input}"

if __name__ == "__main__":
    mcp.run(transport="stdio")

2. Add to your plugin configuration:

const customMCP = new MCPPlugin({
  name: 'custom_mcp',
  servers: [{
    name: 'my_server',
    command: 'python',
    args: ['./mcp_servers/my_server.py'],
    transport: 'stdio'
  }]
});

🐛 Troubleshooting

Common Issues

1. MCP Server Connection Failed

   • Ensure Python dependencies are installed: pip install -r mcp_servers/requirements.txt
   • Check server file permissions: chmod +x mcp_servers/*.py
   • Verify server paths are correct

2. OpenAI API Quota Exceeded

   • Add credits to your OpenAI account
   • Or switch to a different model in the configuration

3. Build Errors

   • Ensure TypeScript is installed: pnpm add -g typescript
   • Clean and rebuild: rm -rf dist && pnpm run build

Debug Mode

Enable debug logging:

# Set debug environment variable
DEBUG=true pnpm start

Logs

Check the console output for:

• MCP server connection status
• Tool loading information
• Agent routing decisions
• Error messages

📚 Documentation

• MCP Plugin Guide - Detailed MCP integration guide
• Agent Configuration - Agent setup and configuration
• Plugin Development - Creating custom plugins

🤝 Contributing

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

📄 License

This project is licensed under the ISC License.

🆘 Support

For issues and questions:

1. Check the troubleshooting section
2. Review the documentation
3. Open an issue on GitHub

---

Happy coding with your AI agents! 🤖✨