LangChain Memory Utility

A reusable LangChain memory utility with support for local file and S3 storage, session-based memory management, and automatic conversation history tracking.

Features

• Flexible Storage: Store memory in local files or AWS S3
• Session-based Memory: Maintain separate memory for each session
• LangChain Integration: Compatible with LangChain's memory interface
• Automatic History: Automatically store and retrieve conversation history
• Easy to Use: Simple API for creating and managing memory

Installation

npm install langchain-memory-util

Quick Start

Local File Storage

const { MemoryStore, SessionMemory } = require('langchain-memory-util');

// Create a local file-based memory store
const store = new MemoryStore({ 
  type: 'local', 
  basePath: './memory' 
});

// Create session memory
const sessionId = 'user123';
const memory = new SessionMemory({ sessionId, store });

// Save conversation context
await memory.saveContext('Hello!', 'Hi there!');
await memory.saveContext('How are you?', 'I am doing well, thank you!');

// Retrieve history
const history = await memory.getHistory();
console.log(history);

S3 Storage

const { MemoryStore, SessionMemory } = require('langchain-memory-util');

// Create an S3-based memory store
const store = new MemoryStore({
  type: 's3',
  bucket: 'my-memory-bucket',
  s3Config: {
    region: 'us-east-1',
    credentials: {
      accessKeyId: 'YOUR_ACCESS_KEY',
      secretAccessKey: 'YOUR_SECRET_KEY'
    }
  }
});

const memory = new SessionMemory({ sessionId: 'user123', store });

With LangChain LLM

const { MemoryStore, SessionMemory } = require('langchain-memory-util');
const { ChatOpenAI } = require('@langchain/openai');
const { ConversationChain } = require('langchain/chains');

// Setup memory
const store = new MemoryStore({ type: 'local', basePath: './memory' });
const memory = new SessionMemory({ sessionId: 'chat1', store });

// Create LLM with memory
const llm = new ChatOpenAI({
  openAIApiKey: process.env.OPENAI_API_KEY,
  temperature: 0.7
});

const chain = new ConversationChain({ llm, memory });

// Conversations are automatically stored
const response = await chain.call({ input: 'Hello, how are you?' });
console.log(response.response);

// Retrieve full conversation history
const history = await memory.getHistory();
console.log('Full conversation:', history);

API Reference

MemoryStore

Constructor

new MemoryStore(config)

Config Options:

• type (string): 'local' or 's3' (default: 'local')
• basePath (string): Directory path for local storage (default: './memory')
• bucket (string): S3 bucket name (required for S3 type)
• s3Config (object): AWS S3 configuration object (required for S3 type)

Methods

• save(sessionId, data): Save data for a session
• load(sessionId): Load data for a session

SessionMemory

Constructor

new SessionMemory({ sessionId, store })

Parameters:

• sessionId (string): Unique identifier for the session
• store (MemoryStore): Memory store instance

Methods

• saveContext(input, output): Save a conversation turn
• loadMemoryVariables(): Load memory variables (LangChain compatible)
• getHistory(): Retrieve full conversation history
• init(): Initialize memory (called automatically)

Examples

Run the included examples:

# Local storage example
npm test

# LLM integration example
npm run example

License

MIT