memory_basic

Memory Basic Example

This example demonstrates how to integrate LangChain's memory functionality with LangGraphGo.

Overview

The memory integration allows you to maintain conversation context across multiple interactions. This example shows various memory types and usage patterns:

1. ConversationBuffer: Stores all conversation history
2. ConversationWindowBuffer: Keeps only the last N conversation turns
3. ConversationTokenBuffer: Maintains history within a token limit
4. ChatMessageHistory: Direct message history management
5. Custom Memory Keys: Using custom input/output/memory keys

Features

• ✅ Multiple memory types (Buffer, Window, Token)
• ✅ Integration with LangGraph
• ✅ Custom memory configuration
• ✅ Message history management
• ✅ Context preservation across turns

Usage

cd examples/memory_basic
go run main.go

Code Highlights

Basic ConversationBuffer

// Create a conversation buffer memory
mem := prebuilt.NewConversationBufferMemory(
    memory.WithReturnMessages(true),
)

// Save conversation context
mem.SaveContext(ctx, map[string]any{
    "input": "Hello, my name is Alice",
}, map[string]any{
    "output": "Hi Alice! Nice to meet you.",
})

// Get all messages
messages, _ := mem.GetMessages(ctx)

ConversationWindowBuffer

// Keep only the last 2 conversation turns
mem := prebuilt.NewConversationWindowBufferMemory(2,
    memory.WithReturnMessages(true),
)

ChatMessageHistory

history := prebuilt.NewChatMessageHistory()

// Add different types of messages
history.AddMessage(ctx, llms.SystemChatMessage{
    Content: "You are a helpful assistant.",
})
history.AddUserMessage(ctx, "Hello!")
history.AddAIMessage(ctx, "Hi! How can I help you?")

// Get all messages
messages, _ := history.Messages(ctx)

Custom Memory Keys

mem := prebuilt.NewConversationBufferMemory(
    memory.WithInputKey("user_input"),
    memory.WithOutputKey("ai_response"),
    memory.WithMemoryKey("chat_history"),
    memory.WithReturnMessages(true),
)

Integration with LangGraph

// Create memory
mem := prebuilt.NewConversationBufferMemory(
    memory.WithReturnMessages(true),
)

// In your graph node
g.AddNode("chat", func(ctx context.Context, state *State) (*State, error) {
    // Load memory
    memVars, _ := mem.LoadMemoryVariables(ctx, map[string]any{})
    
    // Get historical messages
    var historyMessages []llms.ChatMessage
    if history, ok := memVars["history"]; ok {
        if msgs, ok := history.([]llms.ChatMessage); ok {
            historyMessages = msgs
        }
    }
    
    // Use history in your LLM call
    allMessages := append(historyMessages, llms.HumanChatMessage{
        Content: state.Input,
    })
    
    response, _ := llm.GenerateContent(ctx, allMessages)
    
    // Save to memory
    mem.SaveContext(ctx, map[string]any{
        "input": state.Input,
    }, map[string]any{
        "output": response.Choices[0].Content,
    })
    
    return state, nil
})

Memory Types Comparison

API Reference

LangChainMemory

type Memory interface {
    SaveContext(ctx context.Context, inputValues, outputValues map[string]any) error
    LoadMemoryVariables(ctx context.Context, inputs map[string]any) (map[string]any, error)
    Clear(ctx context.Context) error
    GetMessages(ctx context.Context) ([]llms.ChatMessage, error)
}

ChatMessageHistory

type ChatMessageHistory struct {
    // Methods
    AddMessage(ctx context.Context, message llms.ChatMessage) error
    AddUserMessage(ctx context.Context, message string) error
    AddAIMessage(ctx context.Context, message string) error
    Messages(ctx context.Context) ([]llms.ChatMessage, error)
    Clear(ctx context.Context) error
    SetMessages(ctx context.Context, messages []llms.ChatMessage) error
}

Learn More

• LangChain Memory Documentation
• LangGraphGo Documentation
• Memory Chatbot Example