@aigne/did-space-memory

@aigne/did-space-memory

DID Spaces memory system component for AIGNE Framework, providing cloud-based memory storage capabilities with DID Spaces.

Introduction

@aigne/did-space-memory is the DID Spaces memory system component of AIGNE Framework, providing cloud-based memory storage and retrieval functionality using DID Spaces. This component uses decentralized storage services to provide secure and reliable memory persistence capabilities for AI applications.

Features

• Cloud Storage: Uses DID Spaces for secure cloud-based memory storage
• Decentralized: Based on decentralized identity and storage technologies
• Automatic Management: Supports automatic memory file management and README generation
• Secure Authentication: Supports multiple authentication methods to ensure data security
• YAML Format: Uses YAML format for memory storage, making it easy to read and maintain
• TypeScript Support: Complete type definitions providing an excellent development experience

Installation

Using npm

npm install @aigne/did-space-memory

Using yarn

yarn add @aigne/did-space-memory

Using pnpm

pnpm add @aigne/did-space-memory

Basic Usage

import { AIAgent, AIGNE } from "@aigne/core";
import { DIDSpacesMemory } from "@aigne/did-space-memory";
import { OpenAIChatModel } from "@aigne/openai";

// Create AI model instance
const model = new OpenAIChatModel({
  apiKey: process.env.OPENAI_API_KEY,
  model: "gpt-4-turbo",
});

// Create DID Spaces memory system
const memory = new DIDSpacesMemory({
  url: process.env.DID_SPACES_URL!,
  auth: {
    authorization: process.env.DID_SPACES_AUTHORIZATION!,
  },
});

// Create AI agent with cloud-based memory
const agent = AIAgent.from({
  name: "CloudAssistant",
  instructions: "You are a helpful assistant with cloud-based memory.",
  memory: memory,
  inputKey: "message",
});

// Use AIGNE execution engine
const aigne = new AIGNE({ model });

// Invoke agent
const userAgent = await aigne.invoke(agent);

// Send message
const response = await userAgent.invoke({
  message: "I'm John, a doctor, and I like Bitcoin",
});

console.log(response.message);

// Query memory
const response2 = await userAgent.invoke({
  message: "What is my profession?",
});

console.log(response2.message);

Advanced Configuration

Custom Memory Agent Options

import { DIDSpacesMemory } from "@aigne/did-space-memory";

const memory = new DIDSpacesMemory({
  url: process.env.DID_SPACES_URL!,
  auth: {
    authorization: process.env.DID_SPACES_AUTHORIZATION!,
  },
  // Custom memory retriever agent configuration
  retrieverOptions: {
    name: "CustomRetriever",
    instructions: "Custom retrieval instructions",
  },
  // Custom memory recorder agent configuration
  recorderOptions: {
    name: "CustomRecorder",
    instructions: "Custom recording instructions",
  },
  // Other agent configurations
  autoUpdate: true,
});

Using Different Authentication Methods

const memory1 = new DIDSpacesMemory({
  url: "https://your-did-spaces-url.com",
  auth: {
    accessToken: process.env.DID_SPACES_ACCESS_TOKEN!,
  },
});

// Using authorization header authentication
const memory2 = new DIDSpacesMemory({
  url: "https://your-did-spaces-url.com",
  auth: {
    authorization: process.env.DID_SPACES_AUTHORIZATION!,
  },
});

Memory Management Features

const client = memory.client;

// List all memory files
const memories = await client.listObjects({
  prefix: "/memories/",
});

// Read specific memory file
const memoryContent = await client.getObject({
  key: "/memories/conversation-123/memory.yaml",
});

Environment Variables Configuration

Create a .env file and configure the following variables:

DID_SPACES_URL=https://your-did-spaces-url.com
DID_SPACES_AUTHORIZATION=your-authorization-token
# Or use access token
DID_SPACES_ACCESS_TOKEN=your-access-token

License

Elastic-2.0