Codex Mastery
The full Codex course — five modules, twenty lessons, from your first session to shipping a feature with an agent at the keyboard.
Start with 1.1 What Codex Actually IsModule 1 — Foundations
- 1.1 What Codex Actually IsHere's a test. Open a terminal, cd into a real project, type codex, and ask: "Add rate limiting to the login endpoint." If your mental model is "ChatGPT that happens to run in a terminal," you're expecting it to describe an approach and hand you a snippet to paste in yourself. That's not what happens. It reads your routing layer to find the endpoint, checks what middleware is already wired in, decides whether you're on Express, Fastify, or something custom, writes the code, runs your test suite against it, and shows you a diff of exactly what changed and why — before you've typed a second message.
- 1.2 Installing & Your First SessionInstallation is genuinely boring, so let's get it out of the way in one paragraph: npm install -g @openai/codex, or grab it via Homebrew or the direct download if you'd rather skip Node entirely, then run codex login. Sign in with your ChatGPT account (recommended — it picks up whatever plan you're already paying for) or an API key if you're wiring Codex into something headless like CI.
- 1.3 AGENTS.md: Repo MemoryIf this course only left you with one habit, this would be the one.
- 1.4 Sandbox & Approval ModesCodex can edit files, run shell commands, install dependencies, and push to branches. That's real power, and real power without a steering mechanism is how you end up with a horror story instead of a productivity win. OpenAI built two layers of control here — most people only ever learn one, then either feel constantly bottlenecked or eventually get burned.
Module 2 — The Core Toolkit
- 2.1 CLI Commands & Slash CommandsYou already know a few commands from Module 1 — /plan, /status, /diff, /compact. Those cover the basics, but the full toolkit is worth knowing cold, because reaching for the right one at the right moment is most of what makes an experienced Codex user faster than a new one on the exact same task.
- 2.2 Plan Mode: Thinking Before DoingLesson 1.4 introduced plan mode as a safety habit. This lesson is about using it as a speed habit too, which is the part people underestimate.
- 2.3 Reviewing Diffs & Managing ContextTwo separate skills live in this lesson, and they compound: reviewing what changed, and managing what Codex is still carrying around in its head while it works.
- 2.4 Choosing Models & Reasoning EffortNot every task deserves the same amount of thinking, and Codex gives you real control over that dial instead of a single one-size-fits-all setting.
Module 3 — Connecting Everything
- 3.1 IDE Extension: VS Code, Cursor, JetBrainsEverything so far has been the terminal, and the terminal is genuinely the right place for a lot of work — but it's not the only place Codex lives, and knowing when to switch surfaces is its own skill.
- 3.2 MCP: Plugging In GitHub, Figma & MoreAt some point you'll catch yourself doing this: pasting an error from Sentry into a Codex session so it can debug it, then pasting a different one from the same dashboard three days later because the context from last time is gone. Or explaining your team's ticket format for the fourth time. That repetition is a signal, and MCP is the fix.
- 3.3 Cloud Tasks: Fire-and-Forget PRsNot everything needs you sitting at a terminal watching it happen, and treating every task that way is its own kind of inefficiency. Cloud tasks hand a piece of work to Codex running in a hosted container instead of your local machine — you describe the task, close your laptop if you want to, and come back to a finished pull request instead of a stalled session waiting on you.
- 3.4 Browser Extension: Beyond the CodebaseNot every task that benefits from an agent lives in a code editor. The browser extension — available for Chrome, Edge, Brave, Opera, and Vivaldi — runs Codex directly against whatever's open in your browser: an admin console, an internal dashboard, a web app you don't have local access to but do have a login for.
Module 4 — Working Like a Pro
- 4.1 Multi-Agent DelegationA single Codex session, everything up to this point, is one agent doing one job at a time, however large that job is. Multi-agent delegation is the step past that: Codex can spin up subtasks and hand them to parallel subagents, each configurable with its own model choice and reasoning level, working on separate pieces of a larger goal at the same time instead of one long sequential thread.
- 4.2 Vision: From Screenshot to ShippedDescribing a UI in words is lossy in a way that most people don't notice until they see the gap. "A card with the title on top and a button in the bottom right" leaves out spacing, alignment, exact colors, and a dozen small decisions a designer actually made on purpose. Codex's vision capability closes that gap by letting you hand it the actual picture instead of a description of the picture.
- 4.3 Codex as Your Automated Code ReviewerEverything so far has been Codex writing code. This lesson is about the other half: Codex reviewing it — yours, a teammate's, or its own — before a human has to.
- 4.4 Skills: Packaging Reusable WorkflowsPicture the difference between a note that says "format currency as $X.XX" and handing someone your actual finance team's full style guide — when to use parentheses for negatives, which currencies get symbols versus codes, how zero-decimal currencies like JPY get handled. The note is a fact. A Skill is the style guide: a whole reasoning process, packaged as a reusable MCP tool, so Codex can pick it up and apply it correctly without you re-explaining it every time.
Module 5 — Becoming Technically Great
- 5.1 What's New in Codex Right NowA tool that ships this fast rewards people who track why a feature landed, not just that a changelog entry exists. Here's what's actually worth internalizing from late 2026, and what each one changes about how you should be working.
- 5.2 Codex Security: Catch Bugs Before ShipThe more autonomous your setup gets, the more a real security pass pays for itself — and the more expensive it is to skip. Codex Security is worth treating as non-negotiable the moment you're running Codex on anything beyond a throwaway project.
- 5.3 Walkthrough: Shipping a FeatureEverything up to this point has been individual tools. This lesson is where they become one loop — pick a real, small-to-medium feature you actually need (an API endpoint, a new page, a migration you've been putting off) and run it through all six steps, in order, on one real task before you move on.
- 5.4 Your Codex Power-User ChecklistOne condensed reference, worth pinning somewhere you'll actually see it again.
