mb-runtime

Complete Deno runtime framework with AI capabilities, KV storage, and microservices support.

Features

• 🚀 HTTP Server - Hono-based with routing and middleware
• 📦 KV Storage - App-scoped Deno KV with TTL support
• 🤖 LangChain AI - Multi-provider (OpenAI, Anthropic, Google, DeepSeek, Azure)
• 🔐 Authentication - Built-in auth middleware
• 🌐 Nacos Integration - Service registry and discovery
• 📝 Logging - Structured logging with context
• 🔧 HTTP Client - Whitelist-based with rate limiting
• 📊 OpenAPI/Swagger - Auto-generated API documentation

Installation

For Deno Projects

// Using npm specifier (recommended)
import { createApp, config } from "npm:[email protected]";

// Or add to import_map.json
{
  "imports": {
    "mb-runtime": "npm:[email protected]",
    "mb-runtime/": "npm:[email protected]/"
  }
}

For npm Registry (traditional)

npm install mb-runtime

Quick Start

Basic HTTP Server

import { serve } from "https://deno.land/std/http/server.ts";
import { createApp, config } from "npm:mb-runtime";

// Create the application
const app = await createApp();

// Start server
serve(app.fetch, { port: config.port });
console.log(`Server running on port ${config.port}`);

Using Tools Directly

import { 
  createKVStore, 
  createHttpClient, 
  createLangChainClient 
} from "npm:mb-runtime";

// KV Storage
const kv = createKVStore('my-app');
await kv.set('user:123', { name: 'Alice' });

// HTTP Client
const http = createHttpClient({
  id: 'my-app',
  permissions: { http: ['api.example.com'] }
});
const response = await http.fetch('https://api.example.com/data');

// LangChain AI
const ai = await createLangChainClient({
  id: 'my-app',
  permissions: {
    langchain: {
      enabled: true,
      models: ['gpt-4o-mini'],
      features: ['llm', 'memory']
    }
  }
});
const answer = await ai.chat('Hello!');

Configuration

Set environment variables:

# Server
PORT=8002
NODE_ENV=production

# Paths (customizable)
APPS_DIR=./apps               # Default: ./apps

# KV Storage
DENO_KV_PATH=./data/kv.db     # Optional, defaults to memory

# AI Providers (choose one or more)
OPENAI_API_KEY=sk-xxx
ANTHROPIC_API_KEY=sk-ant-xxx
GOOGLE_API_KEY=xxx
DEEPSEEK_API_KEY=sk-xxx

# Azure
AZURE_OPENAI_API_KEY=xxx
AZURE_OPENAI_ENDPOINT=https://xxx

# Nacos (optional)
NACOS_ENABLED=true
NACOS_SERVER_ADDR=nacos:8848

Architecture

mb-runtime/
├── config.js           - Configuration management
├── mod.js             - Main exports
├── http/              - HTTP layer
│   ├── app.js         - Hono app creation
│   ├── middleware/    - Auth, error handling
│   └── routes/        - Route handlers
├── tools/             - Core capabilities ⭐
│   ├── kv.js          - KV storage
│   ├── http.js        - HTTP client
│   └── langchain.js   - AI capabilities
├── services/          - Business services
├── runtime/           - App runtime
└── utils/             - Utilities

API Reference

Core Exports

createApp()

Creates a configured Hono application.

const app = await createApp();

config

Runtime configuration object.

console.log(config.port);  // 8002
console.log(config.paths.apps);  // ./apps

Tools (⭐ AI-Discoverable)

createKVStore(appId: string)

Creates app-scoped KV storage.

Methods:

• get(key) - Get value
• set(key, value, ttl?) - Set value with optional TTL (seconds)
• delete(key) - Delete key
• list(prefix) - List keys with prefix

createHttpClient(appInfo)

Creates whitelist-restricted HTTP client.

Methods:

• fetch(url, options?) - Make HTTP request

createLangChainClient(appInfo, kv?)

Creates LangChain AI client.

Methods:

• chat(message, options?) - Send chat message
• createChain(type, config?) - Create LangChain chain
• saveMessage(role, content) - Save to memory
• getMessages() - Get conversation history
• clearMessages() - Clear memory

Services

NacosService

Service registry integration.

import { NacosService } from "npm:mb-runtime";

const nacos = new NacosService(config);
await nacos.register();
nacos.startHeartbeat();

Usage in Your Project

Project Structure

your-project/
├── main.js           - Your entry point
├── deno.json         - Import map
└── import_map.json   - npm:mb-runtime mapping

Example main.js

import { serve } from "https://deno.land/std/http/server.ts";
import { createApp, config, logger } from "npm:mb-runtime";

logger.info('Starting application...');

const app = await createApp();

serve(app.fetch, {
  port: config.port,
  onListen: ({ port }) => {
    logger.info(`Server listening on port ${port}`);
  }
});

License

MIT

Contributing

Issues and PRs welcome at https://github.com/yourusername/mb-runtime