HazelJSHazelJS
An AI-native backend framework with Agent Runtime, Agentic RAG, and persistent memory built in — designed for stateful AI agents and LLM-powered applications
Why Choose HazelJS?
Built for AI-native applications. Production-grade Agent Runtime, Agentic RAG, and persistent memory—everything you need to build intelligent, stateful AI agents.
AI-Native Architecture
Built from the ground up for AI applications. Not a framework with AI bolted on—AI is at the core of every design decision.
Production Agent Runtime
Stateful execution engine for AI agents. Unlike stateless handlers, agents maintain context, execute tools safely, and support pause/resume workflows.
Agentic RAG
Self-improving retrieval system that plans queries, reflects on results, and adapts strategies. Build intelligent search that learns and improves over time.
Decorator-Based API
Declarative programming with decorators. @Agent, @Tool, @SemanticSearch—add AI capabilities with simple decorators. Clean, type-safe, and intuitive.
Tool Orchestration
Safe, auditable tool execution with approval workflows. Human-in-the-loop for sensitive operations with built-in timeout and retry logic.
Enterprise-Grade Framework
Type-safe, dependency injection, advanced routing, and comprehensive testing. All the framework features you need, built for AI applications.
Production-Ready Security
HazelJS includes enterprise-grade security middleware out of the box. No configuration needed—secure by default.
Security Headers
X-Content-Type-Options, X-Frame-Options, X-XSS-Protection, and HSTS headers automatically applied.
Rate Limiting
Protect your APIs from abuse with configurable rate limiting middleware. In-memory or Redis-backed.
CSRF Protection
Built-in CSRF token generation and validation to protect against cross-site request forgery attacks.
Input Sanitization
Automatic XSS prevention with HTML and string sanitization utilities. Protect against injection attacks.
Built-in Integrations
HazelJS comes with powerful integrations and features out of the box.
Monorepo Packages
11+ modular packages: core, ai, agent, rag, cache, websocket, serverless, prisma, auth, config, swagger, cron, and cli. Install only what you need.
Enterprise Security
Built-in security middleware: Security headers, rate limiting, CSRF protection, and input sanitization. Production-ready by default.
Prisma Integration
First-class ORM support with repository pattern, automatic migrations, and type-safe database access.
Swagger Documentation
Automatic OpenAPI documentation generation with interactive API explorer using swagger-ui-express.
AI Service
Built-in AI integration with OpenAI, Anthropic, Gemini, Cohere, and Ollama. Build AI-powered applications with ease.
Agent Runtime
Production-grade execution engine for stateful AI agents. Built-in tool orchestration, memory integration, and human-in-the-loop workflows.
WebSocket & SSE
Real-time communication with WebSocket gateways, Server-Sent Events, and room management for scalable real-time apps.
RAG & Agentic RAG
Comprehensive RAG with 5 vector stores, Agentic RAG for self-improving retrieval, hybrid search, and semantic embeddings. Build intelligent search systems.
Persistent Memory
Conversation tracking, entity memory, fact storage, and working memory. Build context-aware AI applications that remember and learn.
Latest Blog Posts
Stay up to date with the latest news, tutorials, and updates from the HazelJS team.
Production Agent Runtime: Build Stateful AI Agents That Actually Work
Stop building chatbots. Start building production-grade AI agents with stateful execution, tool orchestration, memory integration, and human-in-the-loop workflows. See real examples of support agents, sales agents, and multi-agent systems.
Beyond Chatbots: Building Self-Improving AI Agents with Agentic RAG
Discover how Agentic RAG and Production Agent Runtime transform AI from simple chatbots into autonomous, self-improving agents that plan, reflect, adapt, and learn—all with a few decorators.
Building Context-Aware AI with the HazelJS Memory System
Discover how to build intelligent, context-aware AI applications with conversation tracking, entity memory, semantic search, and RAG integration using the new Memory System in @hazeljs/rag.
Write Less, Do More
HazelJS makes it easy to write clean, efficient code with modern JavaScript features and built-in optimizations.
Monorepo Structure
Modular packages you can install independently. Choose only what you need for your project.
Security First
Built-in security middleware: Security headers, rate limiting, CSRF protection, and input sanitization.
Advanced Routing
Wildcards, optional params, API versioning with URI, Header, and Media Type strategies.
Prisma Integration
First-class ORM support with repository pattern and automatic database migrations.
// Install only what you need
npm install @hazeljs/core @hazeljs/ai
import { Controller, Get, Injectable } from '@hazeljs/core';
import { AIService } from '@hazeljs/ai';
@Injectable()
class UserService {
constructor(private ai: AIService) {}
async getUsers() {
// Use AI service, Prisma, Cache, etc.
return [{ id: 1, name: 'John' }];
}
}
@Controller('/users')
class UserController {
constructor(private service: UserService) {}
@Get()
getAll() {
return this.service.getUsers();
}
}Loved by Developers
See what developers are saying about HazelJS
The decorator-based API and dependency injection make building scalable applications a breeze. The modular architecture is exactly what we needed!
Built-in AI integration saved us weeks of development time. The testing utilities and security features are fantastic too!
Lightweight and powerful with all the features I need. The Prisma integration, exception filters, and monorepo structure are game-changers!
Join the HazelJS Community
Connect with developers from universities around the world. Share knowledge, collaborate on projects, and build amazing applications together in our invite-only community platform.
University Groups
Join groups from your university. Connect with fellow students, researchers, and faculty members.
Discussions
Share knowledge, ask questions, and engage in meaningful discussions with the community.
Secure & Private
Invite-only access ensures a safe, focused environment for academic and professional collaboration.
Ready to Get Started?
Join the growing community of HazelJS developers and start building amazing applications today.
Build AI-Powered Apps in Minutes
Native OpenAI and Ollama integration. Production-grade Agent Runtime for stateful AI agents. No complex setup, no external libraries—just pure AI power built right into the framework.
OpenAI Ready
GPT-4, GPT-3.5, and DALL-E integration with just a few lines of code. Build chatbots, content generators, and more.
Ollama Support
Run local LLMs with Ollama. Privacy-first AI with Llama 2, Mistral, and other open-source models.
Agent Runtime
Production-grade execution engine for stateful AI agents. Tool orchestration, memory integration, and pause/resume workflows built-in.
As Easy As This
Inject the AI service into any controller or service and start building intelligent features. No configuration needed.
@Injectable()
class ChatService {
constructor(private ai: AIService) {}
async chat(message: string) {
return this.ai.openai.chat({
model: 'gpt-4',
messages: [{ role: 'user', content: message }]
});
}
}Deploy in Seconds, Not Hours
The fastest way to deploy your HazelJS applications. Push to GitHub or Bitbucket and watch your app go live instantly—zero configuration, zero hassle.
Lightning Fast
Deploy in under 30 seconds with optimized build pipelines
Zero Config
No setup required—just connect your repo and deploy
Global Edge
Deployed to edge locations worldwide for ultra-low latency
Auto Previews
Preview every PR automatically before merging
Seamless Git Integration
Connect your GitHub or Bitbucket repository and deploy automatically on every push. It's that simple.
GitHub
BitbucketJoin 9+ developers on the waitlist
Build Intelligent AI Applications
HazelJS includes comprehensive RAG capabilities with Agentic RAG for self-improving retrieval, semantic search, and vector databases. Build smarter applications with autonomous, adaptive AI.
5 Vector Stores
Memory, Pinecone, Qdrant, Weaviate, and ChromaDB with unified interface. Switch seamlessly between dev and production.
Agentic RAG
Self-improving retrieval with query planning, self-reflection, adaptive strategies, and multi-hop reasoning. Autonomous, intelligent search that learns and adapts.
Smart Embeddings
OpenAI and Cohere providers with extensibility. High-quality vector embeddings for semantic search.
Advanced Retrieval
Hybrid search, multi-query retrieval, and BM25 ranking. Combine semantic understanding with keyword precision. Decorator-based API for clean integration.
import { AgenticRAGService } from '@hazeljs/rag/agentic';
import { MemoryVectorStore } from '@hazeljs/rag';
import { OpenAIEmbeddings } from '@hazeljs/ai';
// Initialize with Agentic RAG
const vectorStore = new MemoryVectorStore(new OpenAIEmbeddings());
const agenticRAG = new AgenticRAGService({ vectorStore });
// Self-improving retrieval with query planning,
// self-reflection, and adaptive strategies
const results = await agenticRAG.retrieve('What is HazelJS?');Sponsors & Partners
We're grateful to our sponsors and partners who support the HazelJS community and help us build the future of JavaScript.
Interested in becoming a sponsor or partner?
Get in Touch