MemPro MCP Server

An MCP (Model Context Protocol) server for analyzing MemPro memory profiling data and detecting memory errors using AI-powered analysis.

Overview

This server provides intelligent analysis of MemPro JSON exports, enabling AI assistants to quickly identify memory leaks, fragmentation issues, and large allocations in C++ applications.

Features

MCP Tools

1. analyze_leaks - Analyzes memory leaks and returns prioritized issues

   • Input: json_path (optional, defaults to test_memory_analysis.json)
   • Output: JSON array of memory leak issues with severity, descriptions, and suggestions

2. get_summary - Provides comprehensive memory usage summary

   • Input: json_path (optional)
   • Output: Text summary with key metrics and critical findings

3. get_top_leakers - Returns top N functions causing memory leaks

   • Input: json_path (optional), count (default: 10)
   • Output: Formatted list of top leakers with details

4. analyze_fragmentation - Analyzes memory fragmentation

   • Input: json_path (optional)
   • Output: Fragmentation issues and recommendations

5. find_large_allocations - Identifies unusually large allocations

   • Input: json_path (optional)
   • Output: List of large allocation issues

6. get_all_issues - Comprehensive analysis of all issues

   • Input: json_path (optional)
   • Output: Complete analysis including summary, leaks, fragmentation, and large allocations

MCP Resources

• mempro://stats - Quick access to memory statistics in JSON format

Installation

1. Ensure Go 1.22+ is installed

2. Navigate to the MemProMCP directory:

   cd "C:\Program Files\PureDevSoftware\MemPro\MemProReader\MemProMCP"

3. Download dependencies:

   go mod tidy

4. Build the server:

   go build -o mempro-mcp.exe

Usage

Running the Server

The server uses stdio transport for communication with MCP clients:

./mempro-mcp.exe

Integration with Claude Desktop

Add to your Claude Desktop configuration (claude_desktop_config.json):

{
  "mcpServers": {
    "mempro": {
      "command": "C:\\Program Files\\PureDevSoftware\\MemPro\\MemProReader\\MemProMCP\\mempro-mcp.exe",
      "env": {
        "MEMPRO_JSON_PATH": "C:\\Program Files\\PureDevSoftware\\MemPro\\MemProReader\\test_memory_analysis.json"
      }
    }
  }
}

Environment Variables

• MEMPRO_JSON_PATH - Default path to MemPro JSON file (optional)

Analysis Capabilities

Memory Leak Detection

The analyzer identifies memory leaks based on:

• Leak size and count
• Leak score from MemPro
• Suspect flags
• Function context and call stacks

Severity levels:

• Critical: Suspect leaks > 100KB
• High: Suspect leaks or leaks > 50KB
• Medium: Leaks > 10KB or > 100 allocations
• Low: Other leaks

Intelligent Suggestions

The analyzer provides context-aware suggestions:

• Unknown functions: Enable debug symbols
• STL containers: Check destructors and circular references
• Main function: Review allocation ownership, use RAII
• General: Use smart pointers (std::unique_ptr, std::shared_ptr)

Fragmentation Analysis

Detects memory fragmentation issues:

• High (>80%): Severe fragmentation requiring immediate attention
• Medium (>50%): Moderate fragmentation to monitor

Recommendations include object pooling, memory arenas, and allocation pattern optimization.

Large Allocation Detection

Identifies allocations that may benefit from optimization:

• Average size > 10KB
• Maximum size > 50KB (Medium) or > 100KB (High)

Suggests chunking, streaming, or incremental allocation strategies.

Example Queries for AI

When using this server with an AI assistant:

1. "What are the most critical memory leaks in the application?"
2. "Analyze memory fragmentation and provide recommendations"
3. "Show me the top 5 functions causing memory leaks"
4. "Give me a comprehensive analysis of all memory issues"
5. "What large allocations should I optimize?"

Data Structure

The server parses MemPro JSON with the following key sections:

• CallTrees: Hierarchical allocation call trees
• Functions: Function-level allocation statistics
• Leaks: Detected memory leaks with suspect flags
• PageViews: Memory page usage information
• Types: Allocation type statistics

Development

Project Structure

MemProMCP/
├── main.go       # MCP server setup and tool handlers
├── analyzer.go   # Memory analysis logic
├── types.go      # Data structures for MemPro JSON
├── go.mod        # Go module definition
└── README.md     # This file

Adding New Tools

1. Define tool in setupTools() in main.go
2. Create handler function following the pattern
3. Implement analysis logic in analyzer.go
4. Update README with tool documentation

License

This tool is designed to work with PureDevSoftware's MemPro memory profiler.

Support

For issues or questions about MemPro, visit: https://www.puredevsoftware.com/mempro/

For MCP protocol information: https://modelcontextprotocol.io/