Forge MCP - AI-Augmented Mining Operations Platform

An MCP (Model Context Protocol) server that provides Claude with persistent memory, context management, and autonomous development capabilities. Built for mining professionals to achieve superhuman performance through AI augmentation.

🚀 Current Status (May 28, 2025)

✅ Completed Components

1. Command Centre - Core file operations and system control
2. Memory System - PostgreSQL-based persistent memory with smart retrieval
3. Context Management - Activity-based context loading with token budgeting
4. Development Pipeline - Tool tracking and workflow management

🏗️ Architecture

forge_mcp/
├── mcp_server.py          # Main MCP server
├── tools/                 # MCP tool implementations
│   ├── command_centre.py  # File operations
│   └── memory_tools.py    # Memory operations
├── memory_system/         # Persistent memory
│   ├── memory_manager.py  # Core memory logic
│   ├── context_manager.py # Smart context loading
│   └── schema.sql        # PostgreSQL schema
├── pipeline/             # Development pipeline
│   ├── pipeline_status.yaml # Tool tracking
│   └── view_pipeline.py    # Pipeline viewer
└── router/               # Request routing (future)

🧠 Memory System

Features

• PostgreSQL-based persistent storage
• Hierarchical tagging system
• Smart deduplication
• Context-aware retrieval
• Automatic compression for old memories

Memory Types

• knowledge - Facts and information
• pattern - User preferences and behaviors
• decision - Important choices made
• solution - Working solutions to problems
• error - Issues encountered
• conversation - Discussion summaries

Usage

{
  "action": "commit_memory",
  "parameters": {
    "content": "Important information",
    "memory_type": "knowledge",
    "importance": 8,
    "tags": ["project", "milestone"]
  }
}

🎯 Context Management

Activity-Based Loading

Different activities load different memory contexts:

• brainstorm - Light context (5 memories, patterns/decisions)
• coding - Technical context (15 memories, solutions/errors)
• contract_analysis - Heavy context (20 memories, domain-specific)

Context Budget

• 80,000 tokens working memory
• 20,000 tokens reserved buffer
• Automatic compression when 70% full

📋 Development Pipeline

Tool Status Workflow

1. pending - Initial idea
2. brainstorming - Under discussion
3. ideation - Exploring options
4. ready_for_approval - Needs sign-off
5. approved_for_dev - Ready to build
6. in_development - Being built
7. testing - Quality check
8. complete - Integrated

View Pipeline

cd pipeline/
python view_pipeline.py next      # Next 5 tools to work on
python view_pipeline.py priority  # By priority
python view_pipeline.py contract  # Contract tools only

🛠️ Available Tools

Core Tools (Complete)

• File operations (read, write, create, delete)
• Directory navigation and search
• Memory storage and retrieval
• Context management
• Pattern recording

In Pipeline

• Contract Analyzer (immediate priority)
• Priority Intelligence Engine
• Cost Variance Detective
• Vendor Performance Tracker
• Communication Amplifier

🔧 Installation

1. PostgreSQL Setup

brew install postgresql
brew services start postgresql
psql postgres -c "CREATE USER forge WITH PASSWORD 'forge' CREATEDB;"

2. Python Dependencies

pip install -r requirements.txt

3. Initialize Database

cd memory_system/
python init_db.py

4. Configure Claude Desktop

{
  "mcpServers": {
    "forge": {
      "command": "/bin/bash",
      "args": ["/path/to/forge_mcp/start_mcp.sh"]
    }
  }
}

🎮 Usage

Memory Operations

• Store: Use commit_memory tool
• Retrieve: Use retrieve_memories tool
• Check status: Use get_context_summary tool

Pipeline Management

• View tools: Check pipeline/pipeline_status.yaml
• Update status: Edit YAML and track progress
• No coding until status is approved_for_dev!

🚦 Development Philosophy

• Think First, Code Second - Full brainstorming before implementation
• Modular Architecture - Small tools that combine for big impact
• Value-Easy Methodology - High value, easy implementation first
• Stealth Mode - Results speak louder than technology

📚 Documentation

For Project Owners

• USER_GUIDE.md - Practical guide for daily operations
• TOOL_IDEATION_PROMPT.md - Create consistent tool ideas
• PROJECT_MANAGEMENT_SUMMARY.md - Quick reference for management

For Developers

• ARCHITECTURE.md - System design and components
• IMPLEMENTATION_ROADMAP.md - Development phases and sprints
• REFACTORING_PLAN.md - Code improvement strategy

For Claude Desktop

• CLAUDE_KNOWLEDGE/ - Complete project briefing
  • Load these into Claude Desktop's project knowledge
  • Includes setup instructions and modular documentation

Pipeline & Tracking

• pipeline/ - Development pipeline and tools
  • Enhanced pipeline manager with automation
  • Visual dashboard for progress tracking
  • Sprint planning and velocity metrics

📝 Next Steps

1. Load knowledge base into Claude Desktop (see CLAUDE_KNOWLEDGE/README.md)
2. Review USER_GUIDE.md for daily workflows
3. Use TOOL_IDEATION_PROMPT.md for new tool ideas
4. Monitor progress with enhanced pipeline dashboard

---

Project Vision: Transform mining operations through AI augmentation, starting with contract management excellence and scaling to organizational leadership.

Version: 0.3.0 | Last Updated: December 2024 | Status: Active Development