ACAgentic Craft

Foundation~135 minHazelJS

Agent Foundations

Beginner foundations on HazelJS: what agentic AI means, how AgentRuntime differs from chat and workflows, when not to use an agent, and the runtime mental model.

Authors
editorial-team
Published
Last reviewed

Course promise

Start here if agentic AI is new to you. This track teaches vocabulary and judgment on HazelJS before you write a full production loop.

You will learn what an agent is (and is not), how chatbots and workflows differ, when to stay with a simple prompt or flow, and how AgentRuntime owns a single turn.

Who this is for

  • Engineers who know TypeScript basics but are new to agents
  • Teams aligning on HazelJS Agent OS language
  • Anyone about to start Agentic Development

How to take the course

  1. Read lessons in order — each checkpoint unlocks the next.
  2. Keep a sticky note: goal / tools / done check / non-goals for one real feature.
  3. Prefer understanding over installing every package on day one.
  4. When finished, continue to the builder track (Ops Desk).

Lesson map

#LessonYou leave with
01Agentic AI in plain languageShared vocabulary + sticky-note one-pager
02What is an agent?HazelJS @Agent / @Tool / execute definition
03Agent vs workflow vs chatbotDecision checklist + worked examples
04When not to use an agentLeft-to-right rule + premature smells
05AgentRuntime mental modelPhase-by-phase execute walkthrough

Docs: Agent package, Agent OS.

Learning outcomes

  • Explain agentic AI and HazelJS agents in plain language
  • Choose among @hazeljs/ai, @hazeljs/flow, and AgentRuntime with a checklist
  • Walk one AgentRuntime.execute turn and know when not to use an agent

Lessons

  1. 01Agentic AI in Plain Language
    What “agentic” means for beginners: goals, tools, loops, and why production agents need more than a chat box—mapped to HazelJS.
  2. 02What Is an Agent?
    Define a production agent through HazelJS AgentRuntime: goal loop, tools, observation, and budgets—explained slowly for beginners.
  3. 03Agent vs Workflow vs Chatbot
    Choose among HazelJS chatbots (@hazeljs/ai), deterministic flows (@hazeljs/flow), and AgentRuntime tool loops—with beginner examples and a decision checklist.
  4. 04When Not to Use an Agent
    Default left on the HazelJS spectrum: when a prompt or workflow is enough, premature AgentRuntime smells, and signals that truly justify an open tool loop.
  5. 05AgentRuntime Mental Model
    How AgentRuntime owns context assembly, tool validation, budgets, checkpoints, and stop reasons—step by step for beginners inside a HazelJS app.

Capstone: One-pager for a real feature: goal, done check, chosen HazelJS shape (ai / flow / agent / hybrid), and why the left option fails.

Related patterns