@hazeljs/messaging

@hazeljs/messaging

Multichannel messaging for HazelJS: WhatsApp, Telegram, Viber with LLM-powered bot responses.

Features

• Channel adapters – Telegram (Telegraf), WhatsApp Cloud API, Viber (optional)
• Unified message format – Channel-agnostic IncomingMessage / OutgoingMessage
• LLM integration – Uses @hazeljs/ai providers (OpenAI, Anthropic, etc.) for conversational responses
• Conversation context – Memory (dev) or Redis (production, horizontally scalable)
• Kafka processing – Optional async flow for horizontally scalable message handling
• Webhook controller – Single endpoint per channel for incoming messages

Installation

npm install @hazeljs/messaging @hazeljs/ai @hazeljs/core

For Redis context (horizontal scaling):

npm install ioredis

For Kafka async processing (horizontal scaling):

npm install @hazeljs/kafka

For Viber support:

npm install viber-bot

Quick Start

import { HazelApp } from '@hazeljs/core';
import { MessagingModule } from '@hazeljs/messaging';
import { OpenAIProvider } from '@hazeljs/ai';

const app = new HazelApp({
  imports: [
    MessagingModule.forRoot({
      aiProvider: new OpenAIProvider(process.env.OPENAI_API_KEY),
      systemPrompt: 'You are a helpful support assistant. Keep responses concise.',
      model: 'gpt-4o-mini',
      channels: {
        telegram: { botToken: process.env.TELEGRAM_BOT_TOKEN! },
        whatsapp: {
          accessToken: process.env.WHATSAPP_ACCESS_TOKEN!,
          phoneNumberId: process.env.WHATSAPP_PHONE_NUMBER_ID!,
        },
      },
    }),
  ],
});

app.listen(3000);

Scalable Configuration (Redis + Kafka)

For horizontal scalability, use Redis for context and Kafka for async processing:

import Redis from 'ioredis';

MessagingModule.forRoot({
  aiProvider: new OpenAIProvider(),
  channels: { telegram: { botToken: process.env.TELEGRAM_BOT_TOKEN! } },
  redis: {
    host: process.env.REDIS_HOST ?? 'localhost',
    port: parseInt(process.env.REDIS_PORT ?? '6379', 10),
    password: process.env.REDIS_PASSWORD,
    ttlSeconds: 86400, // 24h
  },
  kafka: {
    brokers: (process.env.KAFKA_BROKERS ?? 'localhost:9092').split(','),
  },
});

• Redis: Stores conversation history; any instance can serve any session.
• Kafka: Webhook returns 200 immediately; consumers process messages asynchronously (scale workers independently).

Webhook URLs

Channel	Method	URL
Telegram	POST	/api/messaging/webhook/telegram
WhatsApp	GET	/api/messaging/webhook/whatsapp (verification)
WhatsApp	POST	/api/messaging/webhook/whatsapp
Viber	POST	/api/messaging/webhook/viber

Configuration

MessagingModuleOptions

• aiProvider – IAIProvider from @hazeljs/ai (e.g. OpenAIProvider)
• systemPrompt – System prompt for the LLM
• model – Model name (default: gpt-4o-mini)
• temperature – 0–1 (default: 0.7)
• maxTokens – Max response tokens (default: 500)
• maxContextTurns – Conversation turns to keep (default: 10)
• customHandler – Override LLM with a custom (msg) => string | Promise<string>
• channels – Channel config (see below)

WhatsApp

• Requires WhatsApp Business API access
• Set WHATSAPP_VERIFY_TOKEN in env for webhook verification
• accessToken, phoneNumberId from Meta for Developers

Telegram

• Create a bot via @BotFather
• Set webhook: https://api.telegram.org/bot<token>/setWebhook?url=https://your-domain/api/messaging/webhook/telegram

Viber

• Create a bot on Viber Developers
• Install optional peer: npm install viber-bot

RAG and Agent Integration

RAG-augmented responses

Use a knowledge base to ground responses. Compatible with @hazeljs/rag RAGService:

import { RAGService } from '@hazeljs/rag';
// ... set up RAGService with vector store, embeddings, etc.

MessagingModule.forRoot({
  aiProvider: new OpenAIProvider(),
  ragService: myRAGService,
  ragTopK: 5,
  ragMinScore: 0.5,
  channels: { telegram: { botToken: process.env.TELEGRAM_BOT_TOKEN! } },
});

Agent handler (CSR-style: tools, RAG, external APIs)

Wire your CSRService or AgentRuntime for full control—tools, RAG, external lookups:

import { MessagingModule } from '@hazeljs/messaging';
import { CSRService } from './csr'; // your CSRService like hazeljs-csr-example

const csrService = new CSRService(...);

MessagingModule.forRoot({
  agentHandler: async ({ message, sessionId }) => {
    const result = await csrService.chat(message.text, sessionId, message.userId);
    return { response: result.response, sources: result.sources };
  },
  channels: { telegram: { botToken: process.env.TELEGRAM_BOT_TOKEN! } },
});

The agent handler receives { message, sessionId, conversationTurns } and can:

• Call AgentRuntime.execute() with tools and RAG
• Use external services (orders, inventory, tickets)
• Return string or { response, sources }

Extending

Custom handler

MessagingModule.forRoot({
  customHandler: async (msg) => {
    if (msg.text === '/help') return 'Available commands: /help, /status';
    return 'Use /help for commands.';
  },
});

Custom adapter

Implement IChannelAdapter and register via providers.