Magic Memory Box Go

Universal Memory for AI Models — A conversation context management library for AI assistants and chatbots.

Magic Memory Box Go provides a simple and flexible way to store and manage conversation history with AI models. The library works with any AI service (OpenAI, Replicate, Claude, DeepSeek, etc.) and supports multiple storage backends.

Integrations

• Reptiloid: Seamless integration with Replicate for AI image and text generation.
• Fantasy: Support for multi-provider AI agents with tool calling.
• Planned: Integration with popular go-openai library for ChatGPT models.

---

🎯 Who is this library for?

• AI application developers who need a ready-made conversation context management system
• Chatbot creators wanting to add memory of previous messages
• Researchers working with different AI models
• Anyone who wants to easily integrate "memory" into their AI projects

---

✨ Key Features

🧠 Smart Context Management

• Automatic history length limiting (keep only the last N messages)
• Configurable message lifetime (TTL)
• Role support: System, User, Assistant, Tool

🔄 Flexible Storage Options

• Built-in memory — Fast in-memory operation
• Redis — Distributed storage for production
• Unified interface — Easy switching between different storage backends

🤝 Compatibility with All AI Services

• Ready-made message format for OpenAI, Replicate, Claude, DeepSeek
• Easy integration with any AI service
• Multi-language support (English, Russian, etc.)

---

🚀 Quick Start

Installation

go get github.com/rmay1er/magic-memory-box-go

Simple Example

package main

import (
    "context"
    "time"
    
    "github.com/rmay1er/magic-memory-box-go/memorybox"
)

func main() {
    ctx := context.Background()
    
    // Create in-memory storage
    cache := memorybox.NewCache()
    
    // Configure the memory box
    mb := memorybox.NewMemoryBox(cache, memorybox.MemoryBoxConfig{
        ContextLenSize: 10,     // Keep last 10 messages
        ExpireTime:     2 * time.Hour, // Auto-expire after 2 hours
    })
    
    // Add system message (assistant personality)
    mb.AddRaw(ctx, "user123", memorybox.SystemRole, "You are a helpful assistant")
    
    // User says something
    mb.Tell(ctx, "user123", "Hello! How are you?")
    
    // Get full history to send to AI model
    messages, _ := mb.GetMemories(ctx, "user123")
    
    // Send messages to any AI service...
    // Get response...
    
    // Save AI response to memory
    mb.Remember(ctx, "user123", "Hello! I'm doing great, thank you!")
}

---

📦 Storage Options

1. Built-in Memory (for development and testing)

import "github.com/rmay1er/magic-memory-box-go/memorybox"

cache := memorybox.NewCache()
// Fast, simple, no external dependencies

2. Redis (for production)

import (
    "github.com/go-redis/redis/v8"
    "github.com/rmay1er/magic-memory-box-go/memorybox"
    "github.com/rmay1er/magic-memory-box-go/rdb"
)

client := redis.NewClient(&redis.Options{
    Addr: "localhost:6379",
})

redisAdapter := rdb.NewRedisAdapter(client, "chat:", true)
// Reliable, distributed, with persistence

---

🔗 AI Service Integration

With Replicate (via Reptiloid)

import "github.com/rmay1er/magic-memory-box-go/convert/replicate"

// Get conversation history
messages, _ := mb.GetMemories(ctx, "user123")

// Convert to Replicate format
replicateMessages := convert.ToReplicate(messages)

// Send to AI model...

---

🎮 Use Cases

Chatbot with Memory

Create a bot that remembers the entire conversation history with a user and can reference previous messages.

Multi-user System

Store separate histories for each user with automatic cleanup of old conversations.

A/B Testing Prompts

Easily switch between different system prompts for different user groups.

Long Conversations

Manage lengthy dialogues by automatically trimming oldest messages while preserving context.

---

📊 Why Magic Memory Box?

Feature	Magic Memory Box	DIY Implementation
Ready Context	✅ Automatic	❌ Need to write code
TTL Support	✅ Built-in	❌ Complex to implement
Storage Switching	✅ 2 lines of code	❌ Rewrite logic
AI-ready Format	✅ Ready-to-use	❌ Manual conversion
Role Support	✅ System/User/Assistant	❌ Need to design

---

🛠️ Ready Examples

The repository includes complete working examples:

• example/cache/ — Example with built-in memory
• example/redis/ — Example with Redis for production

Run them to see the library in action immediately!

---

📈 Performance

• Minimal latency — Optimized for real-time use
• Efficient memory usage — Automatic cleanup of old messages
• Scalability — From one user to millions

---

🤝 Community & Support

Found a bug? Have an improvement idea? Want to add a new storage backend?

• Issues — Report problems
• Pull Requests — Suggest improvements
• Discussions — Share usage ideas

---

📄 License

MIT License — Free to use for any purpose.

---

Made with ❤️ for Go beginners exploring AI generation