HN Showcase
Back to browse
GitHub Repository

A 30-minute course to get up to speed on how AI agents actually work

25 starsTypeScript

A 30-minute course to get up to speed on how AI agents work

by ahd94·Mar 18, 2026·2 points·1 comment
Visit ProjectView on HN

AI Analysis

●●SolidCozyNiche Gem

Builds LangChain's AgentExecutor from 19 to 60 lines without the framework.

Strengths
  • •Nine progressive lessons reveal what frameworks abstract—raw HTTP calls and data structures visible.
  • •Runs entirely in browser via Pyodide—no setup, no signup, mock mode works instantly.
  • •Interactive debugger with play/pause/step controls and turn-aware replay of agent traces.
Weaknesses
  • •Educational repos are common—dozens of 'build an agent from scratch' tutorials exist.
  • •Limited to Anthropic tool_use protocol—doesn't cover OpenAI functions or other standards.
Category
Education
Target Audience

Developers learning how AI agents work under the hood

Similar To

LangChain · CrewAI · AutoGen

Post Description

9 short lessons, ~60 lines of Python. Each adds one concept — tool calling, memory, state, policy gates, self-scheduling — until you see the full agent stack.

The core loop is just a while loop. That's AgentExecutor. That's what the frameworks do.

Runs in the browser (Pyodide). No setup, no signup. Mock mode works instantly, or plug in a free Groq API key for live LLM responses.

Open source: https://github.com/ahumblenerd/tour-of-agents

Similar Projects