Model Context Protocol TypeScript Implementation

Overview

This project implements the Model Context Protocol (MCP) as a practical embodiment of the Personal Intelligence Framework (PIF). Through structured tools and progressive interaction patterns, it creates spaces for meaningful development of understanding between humans and AI.

Quick Start

Prerequisites

• Node.js 18+ and npm
• TypeScript 5.0+
• Model Context Protocol TypeScript SDK
• Claude Desktop Client configured for custom servers

Note: This implementation has been tested on both Windows and macOS/Linux systems.

Setup

1. Clone the repository:

git clone [https://github.com/hungryrobot1/MCP-PIF]
cd mcp-pif

2. Install dependencies:

npm install

3. Configure the server:

   • Configuration is now auto-detected by default, but you can customize:
     • Set the MCP_WORKSPACE_ROOT environment variable to specify a workspace location
     • Or set the MCP_CONFIG environment variable with a JSON string of configuration options
     • Or directly edit src/config.ts to modify the default configuration

4. Build the server:

npm run build

5. Configure Claude Desktop Client:

   • Locate your Claude Desktop Client configuration directory
   • Create or modify claude_desktop_config.json:

     {
         "mcpServers": {
             "mcp-pif": {
                 "command": "node",
                 "args": ["path/to/your/mcp-pif/build/index.js"],
                 "cwd": "path/to/your/mcp-pif",
                 "env": {}
             }
         }
     }

   • Replace path/to/your/mcp-pif with your actual repository path
   • Paths are automatically normalized for your operating system

6. Connect Claude Desktop Client:

   • Start or restart the Claude Desktop Client
   • Select "mcp-pif" as your custom server
   • Start a new chat to begin using the server

Directory Structure

The server will create and manage the following structure in your configured workspace:

workspace/
├── home/
│   ├── meta/
│   │   └── journal/     # For storing journal entries
│   └── projects/        # For user projects

Next Steps

• Review the Module Documentation for available tools
• Explore the Architecture Guide for implementation details
• Check the Development Guide for contribution guidelines

Troubleshooting

• If manually specifying paths, use platform-appropriate separators (backslashes on Windows, forward slashes on macOS/Linux)
• Check the Claude Desktop Client logs if connection fails
• Verify your workspace directory exists and is writable
• Make sure Node.js and TypeScript versions meet requirements

Core Implementation

Available Tools

The implementation provides a set of core tools designed to support structured interaction:

• Filesystem Operations: Navigate and manage workspace context
  • pwd, cd, read, write, mkdir, delete, move, rename
• Reasoning Tools: Create spaces for structured thought
  • reason: Develop connected insights
  • think: Create temporal spaces for contemplation
• Journal System: Maintain framework continuity
  • journal_create: Document developments
  • journal_read: Explore patterns

Basic Usage

// Create a structured thought pattern
reason: {
    thoughts: [
        { content: "Initial observation" },
        {
            content: "Building on previous thought",
            relationType: "sequence",
            relationTo: 0
        }
    ]
}

// Document development
journal_create: {
    title: "Implementation Pattern",
    content: "Insights about development...",
    tags: ["development", "patterns"]
}

Cross-Platform Support

The MCP-PIF server is designed to work seamlessly on Windows, macOS, and Linux environments:

Path Handling

• All file paths are automatically normalized for the current operating system
• The workspace root is detected automatically based on the current environment
• Both absolute and relative paths are supported within the workspace

Configuration

• Environment variables provide a cross-platform way to configure the server
• File operations use Node.js path methods to ensure consistent behavior
• Journal entries and other data are stored in a platform-independent format

Development Workflow

• NPM scripts work on all platforms
• TypeScript compilation produces platform-agnostic JavaScript
• Error handling accounts for platform-specific file system behaviors

Implementation Framework

Module Architecture

The system is built around modular tools that create conditions for structured emergence:

src/
├── core/          # Framework foundations
├── mcp_modules/   # Tool implementations
└── api/           # External integrations

Tool Patterns

Each tool follows consistent patterns while maintaining its unique role:

• Clear interface definitions
• Structured error handling
• State management
• Cross-module interaction

Development Environment

• TypeScript for type safety
• Module-based organization
• Comprehensive logging
• Workspace context management

Theoretical Foundation

Personal Intelligence Framework

The PIF represents a new approach to human-AI collaboration based on:

• Creating conditions for structured emergence
• Maintaining framework-based continuity
• Supporting progressive development
• Enabling meaningful interaction

Structured Emergence

Rather than prescribing fixed patterns, the implementation creates bounded spaces where understanding can emerge through:

• Tool-mediated interaction
• Relationship structures
• Temporal spaces
• Progressive development

Framework-Based Continuity

Understanding develops through:

• Structured documentation
• Pattern discovery
• Historical context
• Evolutionary development

Progressive Disclosure

The system supports different levels of engagement:

• Immediate practical usage
• Pattern discovery
• Framework evolution
• Philosophical alignment

Development Paths

Tool User

For those primarily interested in practical implementation:

1. Start with basic tool usage
2. Explore module documentation
3. Develop interaction patterns
4. Discover emerging capabilities

Framework Developer

For those interested in extending the system:

1. Review module architecture
2. Understand tool patterns
3. Implement new capabilities
4. Maintain framework alignment

Theoretical Explorer

For those interested in deeper patterns:

1. Study implementation principles
2. Observe emerging patterns
3. Contribute to framework evolution
4. Develop new understanding

Contributing

This project welcomes contributions that engage with both implementation and theoretical aspects:

• Tool development
• Documentation improvement
• Pattern discovery
• Framework evolution

Documentation

Comprehensive documentation is available:

• Module Documentation: Individual tool documentation
• Architecture Guide: System design and patterns
• Development Guide: Implementation patterns

Future Directions

The project continues to evolve through:

• New tool development
• Pattern discovery
• Framework refinement
• Community engagement

Philosophy

This implementation embodies a view where:

• Understanding emerges through structured interaction
• Tools create spaces for new patterns
• Development itself becomes philosophical inquiry
• Human and AI intelligence co-evolve

Notes on Usage

The system is more than a set of tools - it is a space for exploring how human and AI intelligence can develop through structured interaction. Each session is an opportunity to discover new patterns of understanding and collaboration.