Model Context Protocol (MCP)

MCP is an open protocol that standardizes how applications provide context to LLMs - think of it like USB-C for AI applications. It enables seamless connection between AI models and various data sources/tools.

🔌 Why MCP?

MCP helps build agents and complex workflows on top of LLMs by providing:

• Pre-built integrations for your LLM to plug into
• Flexibility to switch between LLM providers
• Secure data handling best practices
• Standardized interface for AI applications

🏗️ Core Components

flowchart LR
    A[MCP Host] --> B[MCP Client]
    B --> C[Terminal]
    B --> D[Filesystem]
    B --> E[Memory]
    C --> F[Local Data]
    D --> G[Local Files]
    E --> H[Remote APIs]

Loading

1. MCP Hosts: Applications (like Claude Desktop, IDEs) that need AI context
2. MCP Clients: Protocol handlers that manage server connections
3. MCP Servers: Lightweight programs exposing specific capabilities:
   • Terminal Server: Execute commands
   • Filesystem Server: Access local files
   • Memory Server: Persistent data storage
4. Data Sources:
   • Local: Files, databases on your machine
   • Remote: Web APIs and cloud services

🚀 System Overview

flowchart LR
    User --> Client
    Client --> AI[AI Processing]
    Client --> Terminal[Terminal]
    Client --> Filesystem[Filesystem]
    Client --> Memory[Memory]

Loading

Core Components:

• AI Processing: Google Gemini + LangChain for natural language understanding
• Terminal Server: Executes system commands in isolated workspace
• Filesystem Server: Manages file operations
• Memory Server: Stores and retrieves persistent data

Key Features:

• Automatic server startup as needed
• Secure workspace isolation
• Flexible configuration
• Extensible architecture

📂 File Structure

flowchart TD
    A[mcp] --> B[clients]
    A --> C[servers]
    A --> D[workspace]
    
    B --> E[mcp-client]
    E --> F[main.py]
    E --> G[client.py]
    E --> H[config.json]
    E --> I[.env]
    
    C --> J[terminal]
    J --> K[server.py]
    
    D --> L[memory.json]
    D --> M[notes.txt]

Loading

Key Files:

• clients/mcp-client/main.py: Main client entry point
• clients/mcp-client/langchain_mcp_client_wconfig.py: AI integration
• clients/mcp-client/theailanguage_config.json: Server configurations
• clients/mcp-client/.env: Environment variables
• servers/terminal_server/terminal_server.py: Terminal server
• workspace/memory.json: Persistent memory storage
• workspace/notes.txt: System notes

File Type Breakdown:

• Python Files (60%):

  • Core application logic and business rules
  • Server implementations and client applications
  • Includes both synchronous and asynchronous code
  • Follows PEP 8 style guidelines

• JSON Files (20%):

  • Configuration files for servers and services
  • API request/response schemas
  • Persistent data storage format
  • Strict schema validation enforced

• Text Files (15%):

  • System documentation (READMEs, guides)
  • Developer notes and annotations
  • Temporary data storage
  • Plaintext logs and outputs

• Other Formats (5%):

  • Environment files (.env)
  • Git ignore patterns
  • License information
  • Build configuration files

🔌 Client Components

flowchart TD
    A[User Input] --> B[Client]
    B --> C{Type?}
    C -->|Command| D[Terminal]
    C -->|File| E[Filesystem]
    C -->|Memory| F[Storage]
    C -->|AI| G[Gemini]
    D --> H[Response]
    E --> H
    F --> H
    G --> H
    H --> I[Output]

Loading

Main Client Files:

• langchain_mcp_client_wconfig.py: Main client application
• theailanguage_config.json: Server configurations
• .env: Environment variables

Key Features:

• Manages multiple MCP servers
• Integrates Google Gemini for natural language processing
• Handles dynamic response generation
• Processes LangChain objects

Configuration:

1. theailanguage_config.json:

{
  "mcpServers": {
    "terminal_server": {
      "command": "uv",
      "args": ["run", "../../servers/terminal_server/terminal_server.py"]
    },
    "memory": {
      "command": "npx.cmd",
      "args": ["@modelcontextprotocol/server-memory"],
      "env": {"MEMORY_FILE_PATH": "workspace/memory.json"}
    }
  }
}

2. .env Setup:

GOOGLE_API_KEY=your_api_key_here
THEAILANGUAGE_CONFIG=clients/mcp-client/theailanguage_config.json

Setup Steps:

1. Create .env file in clients/mcp-client/
2. Add required variables
3. Restart client after changes

🖥️ Server Components

classDiagram
    class TerminalServer {
        +path: String
        +run()
        +validate() 
        +execute()
    }
    TerminalServer --|> FastMCP
    class FastMCP {
        +decorate()
        +transport()
    }

Loading

Terminal Server

• Purpose: Executes system commands in isolated workspace
• Key Features:
  • Fast command execution
  • Secure workspace isolation
  • Comprehensive logging
• Technical Details:
  • Uses FastMCP for transport
  • Validates commands before execution
  • Captures and returns output

Workspace Files

memory.json

• Purpose: Persistent data storage
• Operations:
  • Store/update/read data
  • Query specific information
• Example Structure:

{
  "user_preferences": {
    "favorite_color": "blue",
    "interests": ["science fiction"]
  },
  "system_state": {
    "last_commands": ["git status", "ls"]
  }
}

notes.txt

• Purpose: System documentation and notes
• Content Types:
  • User documentation (40%)
  • System notes (30%)
  • Temporary data (20%)
  • Other (10%)

🛠️ Local Setup Guide

Prerequisites

• Python 3.9+
• Node.js 16+
• Google API Key
• UV Package Manager

Installation Steps

1. Clone the repository:

   git clone https://github.com/Techiral/mcp.git
   cd mcp

2. Set up Python environment:

   python -m venv venv
   # Linux/Mac:
   source venv/bin/activate
   # Windows:
   venv\Scripts\activate
   pip install -r requirements.txt

3. Configure environment variables:

   echo "GOOGLE_API_KEY=your_key_here" > clients/mcp-client/.env
   echo "THEAILANGUAGE_CONFIG=clients/mcp-client/theailanguage_config.json" >> clients/mcp-client/.env

4. Install Node.js servers:

   npm install -g @modelcontextprotocol/server-memory @modelcontextprotocol/server-filesystem

Verification Checklist:

• Repository cloned
• Python virtual environment created and activated
• Python dependencies installed
• .env file configured
• Node.js servers installed

🚀 Usage Instructions

Basic Usage

1. Start the client:

python clients/mcp-client/langchain_mcp_client_wconfig.py

2. Type natural language requests and receive responses

Command Examples

File Operations:

Create a file named example.txt
Search for "function" in all Python files
Count lines in main.py

Web Content:

Summarize https://example.com
Extract headlines from news site

System Commands:

List files in current directory
Check Python version
Run git status

Memory Operations:

Remember my favorite color is blue
What preferences did I set?
Show recent commands

Server Configuration

Key Configuration Files:

• theailanguage_config.json: Main server configurations
• .env: Environment variables

Example Server Configs:

{
  "terminal_server": {
    "command": "uv",
    "args": ["run", "servers/terminal_server/terminal_server.py"]
  },
  "memory": {
    "command": "npx.cmd",
    "args": ["@modelcontextprotocol/server-memory"],
    "env": {"MEMORY_FILE_PATH": "workspace/memory.json"}
  }
}

Configuration Tips:

• Use absolute paths for reliability
• Set environment variables for sensitive data
• Restart servers after configuration changes

🛠️ Troubleshooting

Common Issues & Solutions:

1. Authentication Problems:

   • Verify Google API key in .env
   • Check key has proper permissions
   • Regenerate key if needed

2. File Operations Failing:

   # Check permissions
   ls -la workspace/
   
   # Restart filesystem server
   npx @modelcontextprotocol/inspector uvx mcp-server-filesystem

3. Memory Operations Failing:

   # Verify memory.json exists
   ls workspace/memory.json
   
   # Restart memory server
   npx @modelcontextprotocol/server-memory

Debugging Tools:

• Enable verbose logging:

  echo "LOG_LEVEL=DEBUG" >> clients/mcp-client/.env

• List running servers:

  npx @modelcontextprotocol/inspector list

Support:

• Documentation
• Report Issues

🤝 How to Contribute

Getting Started:

1. Fork and clone the repository
2. Set up development environment (see Local Setup Guide)

Development Workflow:

# Create feature branch
git checkout -b feature/your-feature

# Make changes following:
# - Python: PEP 8 style
# - JavaScript: StandardJS style
# - Document all new functions

# Run tests
python -m pytest tests/

# Push changes
git push origin feature/your-feature

Pull Requests:

• Reference related issues
• Describe changes clearly
• Include test results
• Squash commits before merging

Code Review:

• Reviews typically within 48 hours
• Address all feedback before merging

Recommended Setup:

• VSCode with Python/JS extensions
• Docker for testing
• Pre-commit hooks