memg 0.5.3

True memory for AI - lightweight, generalist, AI-made, AI-focused

MEMG 🧠⚡

True memory for AI - lightweight, generalist, AI-made, AI-focused

MEMG is an AI-enhanced memory management system that provides intelligent, persistent memory capabilities for AI agents and applications. Built on top of the robust memg-core foundation, MEMG adds sophisticated AI-powered features for enhanced search, intelligent processing, and seamless integration.

🚀 Quick Start

pip install memg

import memg

# Add memories with intelligent processing
memory = memg.add_memory("task", {
    "statement": "Implement user authentication system",
    "details": "Need to add JWT-based auth with role management",
    "priority": "high"
}, user_id="your_user_id")

# AI-enhanced search with natural language
results = memg.search_enhanced("authentication tasks with high priority", user_id="your_user_id")

# Access AI utilities for text generation
ai = memg.GenAI(system_instruction="You are a helpful assistant")
response = ai.generate_text("Analyze this user requirement: implement authentication")

✨ Key Features

🎯 AI-Enhanced Memory

• Intelligent Search: Natural language queries with semantic understanding
• Smart Processing: AI-powered content analysis and categorization
• Enhanced Retrieval: Context-aware memory retrieval with relevance scoring

🏗️ Built on Solid Foundation

• memg-core Integration: Leverages battle-tested core memory management
• Production Ready: Comprehensive testing, quality tools, and CI/CD
• Modern Architecture: Clean separation between core storage and AI enhancements

🔌 Flexible Integration

• MCP Server: Memory Context Protocol server for tool integration
• RESTful API: Standard HTTP endpoints for web applications
• Python SDK: Rich Python API with async/sync support

🛠️ Developer Experience

• Rich Configuration: YAML-based schemas and flexible setup
• Comprehensive Testing: Unit and integration test suites
• Quality Tools: Ruff, MyPy, Bandit for code quality
• Type Safety: Full type hints and runtime validation

📦 Architecture

MEMG Ecosystem
├── memg-core (PyPI)          # Foundation: storage, search, schemas
├── MEMG (this package)       # AI enhancements, processing, utilities
└── integrations/             # MCP server, APIs, external tools

Core Components

• memg.core: Integration layer with memg-core
• memg.ai: AI-powered processing and analysis
• memg.search: Enhanced search orchestration
• memg.api: High-level API interfaces
• memg.utils: Utilities and schema management

🎮 Usage Examples

Basic Memory Operations

import memg

# Create and store memories
memory = memg.add_memory("note", {
    "statement": "API design patterns research",
    "details": "Investigated REST vs GraphQL for user management API",
    "project": "web-app"
}, user_id="your_user_id")

# Search with natural language
results = memg.search("API design research for web applications")

# Enhanced AI-powered search
ai_results = memg.search_enhanced("show me notes about web development", user_id="your_user_id")

AI-Enhanced Processing

from memg import GenAI

# Initialize with system instruction
ai = GenAI(system_instruction="You are a software architecture expert")

# Generate structured JSON responses
schema = {
    "type": "object",
    "properties": {
        "summary": {"type": "string"},
        "key_patterns": {"type": "array", "items": {"type": "string"}},
        "recommendations": {"type": "array", "items": {"type": "string"}}
    },
    "required": ["summary", "key_patterns", "recommendations"]
}

analysis = ai.generate_json(
    "Analyze this document about software architecture patterns...",
    schema
)

# Generate free-text responses
insights = ai.generate_text("Provide insights on these architecture patterns")

MCP Server Integration

# Run the MCP server for tool integration
cd integrations/mcp/
./build_and_run.sh

# Server runs on localhost:8787 with full MEMG capabilities
curl http://localhost:8787/health

🏁 Getting Started

Installation

# Install MEMG with AI features
pip install memg

# For development setup
git clone https://github.com/genovo-ai/memg.git
cd memg
pip install -e ".[dev]"

Configuration

Create a memory configuration:

# config/my_config.yaml
entities:
  task:
    required: [statement]
    optional: [assignee, priority, status, due_date]
  note:
    required: [statement, details]
    optional: [project, tags]

import memg

# Initialize with custom configuration
memory = memg.create_memory_from_yaml("config/my_config.yaml")

Memory Server

For persistent memory with web access:

# Start the MCP server
cd integrations/mcp/
docker-compose up -d

# Or use the convenience script
./build_and_run.sh

🔧 Development

Quality Tools

# Run all quality checks
make quality-check

# Individual tools
make lint          # Ruff linting
make typecheck     # MyPy type checking
make security      # Bandit security scan
make test          # Full test suite

Testing

# Run tests
make test-all

# Fast tests only
make test-fast

# Integration tests
make test-integration

# Coverage report
make test-coverage

🤝 Contributing

We welcome contributions! Please see our development workflow:

1. Fork & Clone: Fork the repository and clone locally
2. Setup: pip install -e ".[dev]" for development dependencies
3. Quality: Run make quality-check before committing
4. Test: Ensure make test-all passes
5. PR: Submit a pull request with clear description

Development Standards

• Code Quality: Ruff formatting and linting
• Type Safety: MyPy type checking required
• Security: Bandit security scanning
• Testing: Comprehensive test coverage
• Documentation: Clear docstrings and examples

📚 Documentation

• API Reference: Coming soon
• Architecture Guide: See src/memg/ for component structure
• Integration Guide: Check integrations/mcp/README.md
• Configuration: Explore config/ directory for examples

🛡️ Security

• Bandit Scanning: Automated security vulnerability detection
• Dependency Management: Regular security updates
• Input Validation: Comprehensive data validation
• Safe Defaults: Secure-by-default configuration

📄 License

MIT License - see LICENSE file for details.

🌟 Related Projects

• memg-core: Foundation memory management system
• MCP Tools: Memory Context Protocol integrations

---

Built with ❤️ by the MEMG Team

True memory for AI - making intelligent agents truly intelligent 🧠✨