Course

How AI Coding Agents Work

You use AI coding tools every day — but what are they actually doing between your prompt and the diff? This course opens the hood on Cursor, Claude Code, Codex, and Windsurf: the read-plan-edit-run loop at their core, how they index a whole repo, the tools they reach for, and the real differences behind the marketing. Six short lessons for developers, every screen drawn, no version numbers to memorize. It's the companion to [vibecoding](glossary://vibecoding) — that's how to *use* these tools; this is how they're *built*.

6 lessonsIntermediate39 min

Start the course Get the app

Course lessons

How a coding agent works

The read-plan-edit-run loop at the heart of every coding agent — and why it retries.

›

How Claude Code works

A terminal-native agent: the same think-act-observe loop, pointed at your codebase.

›

Indexing, planning, editing

How an agent understands a repo it has never seen, then makes a precise change.

›

Tool use in a coding agent

What an agent can actually do — read, edit, run, search — and how MCP adds more.

›

Cursor vs Claude Code vs Codex vs Windsurf

Same core loop, different homes and trade-offs — how to tell them apart without the hype.

›

The whole machine

Pull it together — the one mental model that explains every coding agent.

›

Take the full course

Learn it properly — tracked & personalized

Follow the whole course as a guided journey — sequential lessons, progress, streaks and a personalized path, all in the app.

Get the app