# HazelJS

> AI-native TypeScript backend and Agent OS for production agents — durable runs, Skillgate, policies, DNA, RAG, and APIs in one Node stack. Not a bare-metal OS; not a Python-only kernel.

## Documentation

- [Introduction](https://hazeljs.ai/docs): Overview, architecture, and core philosophy
- [Installation](https://hazeljs.ai/docs/installation): Setup, prerequisites, and project scaffolding
- [Agent OS lander](https://hazeljs.ai/agent-os): What Agent OS is and where to start

## Tutorials

- [Tutorial: Build a REST API](https://hazeljs.ai/docs/guides/tutorial-rest-api): End-to-end guide building a Task Management API with CRUD, validation, DI, modules, guards, and error handling
- [Production AI agents in Node.js](https://hazeljs.ai/docs/guides/production-ai-agents-nodejs): Ship AI-native agents with Agent OS
- [Build a TypeScript RAG API](https://hazeljs.ai/docs/guides/build-typescript-rag-api): RAG API how-to

## Guides

- [Agent OS](https://hazeljs.ai/docs/guides/agent-os): AI-native TypeScript agent lifecycle — durable runs, SQL stores, leases, Skillgate, loops, policies, DNA
- [Skillgate](https://hazeljs.ai/docs/guides/skillgate): OpenAPI / REST → curated, governed agent skills
- [Controllers](https://hazeljs.ai/docs/guides/controllers): Decorator-based routing, request handling, parameter extraction, status codes, headers, redirects
- [Providers](https://hazeljs.ai/docs/guides/providers): Dependency injection, services, scopes (singleton, transient, request), custom providers
- [Modules](https://hazeljs.ai/docs/guides/modules): Module composition, imports/exports, dynamic modules, global modules
- [Guards](https://hazeljs.ai/docs/guides/guards): Authentication, authorization, role-based access control
- [Interceptors](https://hazeljs.ai/docs/guides/interceptors): Request/response transformation, logging, caching
- [Pipes](https://hazeljs.ai/docs/guides/pipes): Data transformation, validation, ParseIntPipe, ValidationPipe
- [Middleware](https://hazeljs.ai/docs/guides/middleware): Timeout, CORS, rate limiting, security headers
- [Exception Filters](https://hazeljs.ai/docs/guides/exception-filters): Error handling, custom exception filters
- [Security](https://hazeljs.ai/docs/guides/security): Security headers, CSRF, rate limiting, input sanitization
- [Memory](https://hazeljs.ai/docs/guides/memory): AI agent memory system, conversation tracking, entity memory
- [Agentic RAG](https://hazeljs.ai/docs/guides/agentic-rag): Self-improving retrieval, query planning, adaptive strategies
- [RAG Patterns](https://hazeljs.ai/docs/guides/rag-patterns): Common RAG patterns and best practices
- [Vector Stores](https://hazeljs.ai/docs/guides/vector-stores): Memory, Pinecone, Qdrant, Weaviate, ChromaDB
- [Serverless Deployment](https://hazeljs.ai/docs/guides/serverless-deployment): AWS Lambda, Vercel, Google Cloud Functions
- [Glossary](https://hazeljs.ai/docs/guides/glossary): Agent OS, AgentRun, lease, Skillgate, and more

## Packages

- [AI (@hazeljs/ai)](https://hazeljs.ai/docs/packages/ai): OpenAI, Anthropic, Gemini, Cohere, Ollama integration, @AITask, @AIFunction, @AIValidate decorators
- [Agent (@hazeljs/agent)](https://hazeljs.ai/docs/packages/agent): Agent OS kernel — durable runs, leases, tools, memory, HITL
- [Agent VM (@hazeljs/agent-vm)](https://hazeljs.ai/docs/packages/agent-vm): Effect-typed tools, speculation, transactional undo
- [Organism (@hazeljs/organism)](https://hazeljs.ai/docs/packages/organism): Mission-defined self-organizing agent societies — genes, environment signals, constitution, economy, survival
- [Agent Gatekeeper (@hazeljs/agent-gatekeeper)](https://hazeljs.ai/docs/packages/agent-gatekeeper): Authorize every tool call before execution
- [Skillgate (@hazeljs/skillgate)](https://hazeljs.ai/docs/packages/skillgate): OpenAPI → governed agent skills for Agent OS / MCP
- [Auth (@hazeljs/auth)](https://hazeljs.ai/docs/packages/auth): JWT authentication, guards, role-based access control
- [Cache (@hazeljs/cache)](https://hazeljs.ai/docs/packages/cache): Multi-tier caching (Memory, Redis, CDN)
- [CLI (@hazeljs/cli)](https://hazeljs.ai/docs/packages/cli): Scaffolding and code generation
- [Config (@hazeljs/config)](https://hazeljs.ai/docs/packages/config): Type-safe configuration with validation
- [Config Server (@hazeljs/config-server)](https://hazeljs.ai/docs/packages/config-server): Git-backed distributed configuration, AES {cipher} secrets, HTTP refresh
- [Cron (@hazeljs/cron)](https://hazeljs.ai/docs/packages/cron): Decorator-based scheduled tasks
- [Discovery (@hazeljs/discovery)](https://hazeljs.ai/docs/packages/discovery): Service discovery, registry, load balancing, health checks, retries
- [Kafka (@hazeljs/kafka)](https://hazeljs.ai/docs/packages/kafka): Event-driven consumers, producers, stream processing pipelines
- [Ops Agent (@hazeljs/ops-agent)](https://hazeljs.ai/docs/packages/ops-agent): AI-powered Jira, Slack, incident triage, DevOps workflows
- [PDF-to-Audio (@hazeljs/pdf-to-audio)](https://hazeljs.ai/docs/packages/pdf-to-audio): Convert PDFs to audio with OpenAI TTS, async jobs, AI summaries
- [Predictive Scaling (@hazeljs/predictive-scaling)](https://hazeljs.ai/docs/packages/predictive-scaling): Forecast traffic and raise Kubernetes HPA before spikes
- [Prisma (@hazeljs/prisma)](https://hazeljs.ai/docs/packages/prisma): Prisma ORM integration, repository pattern
- [RAG (@hazeljs/rag)](https://hazeljs.ai/docs/packages/rag): Vector search, embeddings, retrieval strategies, text splitting
- [Self-Healing (@hazeljs/self-healing)](https://hazeljs.ai/docs/packages/self-healing): Diagnose failures, rollback config, restart pods, boost HPA
- [Serverless (@hazeljs/serverless)](https://hazeljs.ai/docs/packages/serverless): AWS Lambda, Google Cloud Functions adapters
- [Swagger (@hazeljs/swagger)](https://hazeljs.ai/docs/packages/swagger): Automatic OpenAPI/Swagger documentation
- [WebSocket (@hazeljs/websocket)](https://hazeljs.ai/docs/packages/websocket): WebSocket and Server-Sent Events
- [MCP (@hazeljs/mcp)](https://hazeljs.ai/docs/packages/mcp): Export ToolRegistry skills to Cursor / Claude Desktop

## API Reference

- [API Reference](https://hazeljs.ai/docs/api): Complete API reference with all exported symbols

## Blog

- [Production ops stack](https://hazeljs.ai/blog/production-ops-stack-config-healing-scaling): Config server, self-healing, and predictive HPA
- [What is an Agent OS?](https://hazeljs.ai/blog/what-is-agent-os): Definition and disambiguation for TypeScript / Node
- [Agent OS Gamma](https://hazeljs.ai/blog/agent-os-gamma-durable-runs): Durable runs, SQL stores, worker leases
- [Skillgate](https://hazeljs.ai/blog/skillgate-openapi-to-agent-skills): OpenAPI → governed agent skills
- [What's new in Agent OS](https://hazeljs.ai/blog/whats-new-in-agent-os): Lifecycle launch post
- [Blog index](https://hazeljs.ai/blog): Release notes, package deep-dives, tutorials

## Community

- [Build with HazelJS](https://hazeljs.ai/build): Community program — Agent OS challenges, public TypeScript projects, showcase submissions
- [Community Showcase](https://hazeljs.ai/build/showcase): Featured independent HazelJS Agent OS repositories
- [Community Projects](https://hazeljs.ai/build#showcase): Featured projects on the Build page
- [Agent OS Challenges](https://hazeljs.ai/build#challenges): Production-style build challenges (incident, refund, CloudOps, MCP, and more)
- [How projects get featured](https://hazeljs.ai/build#get-featured): Submit → review → featured on the live showcase
- [Challenge season](https://hazeljs.ai/build): Q4 2026 (Oct 1 – Dec 31, 2026) — goal 20 projects / 10 builders

## Optional

- [Full Documentation for LLMs](https://hazeljs.ai/llms-full.txt): Complete documentation in a single text file