@t0ken.ai/memoryx-sdk

MemoryX Node.js SDK - Enable AI Agents with persistent memory

Installation

npm install @t0ken.ai/memoryx-sdk

Quick Start

import { MemoryXSDK, PRESET } from '@t0ken.ai/memoryx-sdk';

// Use preset mode
const sdk = new MemoryXSDK({ preset: 'batch' });

// Add memory
await sdk.addMemory("User prefers Python for data analysis");

// Search memories
const results = await sdk.search("data analysis");
console.log(results.data);

Send Strategies

SDK supports flexible send strategy configuration:

Preset Modes

// Realtime mode - send immediately
const sdk = new MemoryXSDK({ preset: 'realtime' });

// Batch mode - 50 items or 5 seconds
const sdk = new MemoryXSDK({ preset: 'batch' });

// Conversation mode - 2 rounds or 30 minutes
const sdk = new MemoryXSDK({ preset: 'conversation' });

Custom Strategy

const sdk = new MemoryXSDK({
  strategy: {
    rounds: 3,        // Send after 3 conversation rounds
    batchSize: 100,   // Or 100 messages
    intervalMs: 60000 // Or 1 minute
  }
});

Fully Custom

const sdk = new MemoryXSDK({
  strategy: {
    customTrigger: (stats) => {
      // Custom logic
      return stats.rounds >= 2 || stats.totalTokens >= 4000;
    }
  }
});

API

Memory Operations

// Add memory
await sdk.addMemory(content, metadata?);

// Search memories
const results = await sdk.search(query, limit?);

// List memories
const memories = await sdk.list(limit?, offset?);

// Delete memory
await sdk.delete(memoryId);

// Get quota
const quota = await sdk.getQuota();

Conversation Operations

// Add conversation message
await sdk.addMessage('user', 'Hello');
await sdk.addMessage('assistant', 'Hi! How can I help you?');

// Start new conversation
sdk.startNewConversation();

Queue Operations

// Manual flush queue
await sdk.flush();

// Get queue stats
const stats = await sdk.getQueueStats();

Strategy Parameters

| Parameter | Type | Description | |-----------|------|-------------| | rounds | number | Send after N conversation rounds | | batchSize | number | Send after N messages | | intervalMs | number | Send after N milliseconds | | maxTokens | number | Send after N tokens | | customTrigger | function | Custom trigger function |

Queue Stats

interface QueueStats {
  messageCount: number;      // Current message count
  rounds: number;            // Current rounds
  totalTokens: number;       // Total tokens
  oldestMessageAge: number;  // Age of oldest message
  conversationId: string;    // Conversation ID
}

License

MIT