Claude-MCP-Integration

jpgtex/Claude-MCP-Integration

3.2

If you are the rightful owner of Claude-MCP-Integration and would like to certify it and/or have it hosted online, please leave a comment on the right or send an email to henry@mcphub.com.

This project provides a standalone, production-ready Model Context Protocol (MCP) server designed for integration with Claude Desktop on macOS.

šŸ§  Claude Desktop MCP Integration

Standalone, Production-Ready MCP Server for Claude Desktop

Overview

This project provides a lean, stable Model Context Protocol (MCP) server specifically designed for Claude Desktop integration on macOS. It leverages proven components from the Astraeus AI system while maintaining complete independence and focusing on reliability over experimental features.

Key Features

  • šŸ”’ Complete Separation: Independent from Astraeus development - no shared dependencies
  • āš” Proven Components: Only battle-tested memory caching, compression, and GPU acceleration
  • šŸ›”ļø Production Hardened: Comprehensive error handling, memory management, and monitoring
  • šŸš€ Performance Optimized: 120:1 compression ratio with Metal Performance Shaders acceleration
  • šŸ³ Podman Ready: Containerized for reproducibility with Apple Silicon support

Architecture

~/Desktop/Claude-MCP-Integration/
ā”œā”€ā”€ core/                       # Python business logic
ā”‚   ā”œā”€ā”€ interfaces/            # Clean API facades for each service
ā”‚   ā”œā”€ā”€ vendored/              # Copied & adapted Astraeus components
ā”‚   ā””ā”€ā”€ mcp_bridge.py          # MCP protocol handler
ā”œā”€ā”€ mcp/                       # MCP resource and tool definitions
ā”œā”€ā”€ utils/                     # Shared utilities and helpers
ā”œā”€ā”€ tests/                     # Comprehensive test suite
ā”œā”€ā”€ scripts/                   # Build, deployment, and diagnostic tools
ā”œā”€ā”€ podman/                    # Container specifications
ā””ā”€ā”€ docs/                      # Documentation and guides

Core Components

Memory Server

  • LRU Cache: High-performance caching with configurable limits
  • Embedding Search: Vector similarity search for context retrieval
  • Session Persistence: SQLite backing for continuity across restarts

Compression Engine

  • 120:1 Ratio: Neural + zlib hybrid compression pipeline
  • Streaming Support: Handle large files efficiently
  • GPU Acceleration: Metal Performance Shaders on Apple Silicon

Resource Monitoring

  • System Metrics: CPU, RAM, GPU utilization tracking
  • Health Checks: Automated degradation detection and recovery
  • Structured Logging: JSON logs with rotation and filtering

File Operations

  • Cached I/O: Memory-backed file operations with persistence
  • Batch Processing: Efficient handling of multiple files
  • Safe Operations: Atomic writes with backup recovery

Design Principles

  1. Stability First: Only proven, non-experimental components
  2. Memory Safety: Proper cleanup, monitoring, automatic garbage collection
  3. Connection Reliability: Timeout management, retries, circuit breakers
  4. Error Recovery: Graceful degradation and comprehensive diagnostics
  5. Resource Limits: Configurable bounds with automatic enforcement

Installation

cd ~/Desktop/Claude-MCP-Integration
pip install -e .

Configuration

Update your Claude Desktop configuration:

{
  "mcpServers": {
    "astraeus-tools": {
      "command": "python3",
      "args": ["/Users/jpgair/Desktop/Claude-MCP-Integration/core/mcp_bridge.py"],
      "env": {
        "MCP_LOG_LEVEL": "INFO",
        "CACHE_SIZE": "1GB", 
        "GPU_ENABLED": "true",
        "COMPRESSION_LEVEL": "120"
      }
    }
  }
}

Development Status

šŸš§ In Development - Following proven enterprise patterns from Astraeus V4.0 deployment

Requirements

  • Platform: macOS (Apple Silicon preferred for GPU acceleration)
  • Python: 3.13+
  • Memory: 2GB+ recommended for full feature set
  • Storage: 1GB for cache and logs

License

MIT License - See LICENSE file for details

Relationship to Astraeus

This project is completely independent of the Astraeus-1.0-Development codebase. While it leverages proven algorithms and patterns from Astraeus, all code has been:

  • Manually copied and adapted (no symlinks or references)
  • Stripped of Astraeus-specific dependencies
  • Renamed to use claude_mcp.* package structure
  • Simplified for MCP-specific use cases
  • Hardened based on latest MCP ecosystem issue analysis (Sept 2025)

This ensures your Astraeus development remains unaffected while providing Claude Desktop with stable, high-performance tools.

Built for production reliability ā€¢ Optimized for Claude Desktop ā€¢ Powered by proven AI system components