
Explore Codex through a hands-on demo project and master its basic and advanced features. Use the Codex CLI, IDE integration, and the native desktop app to maximize Codex.
Explore codex, the coding-optimized AI tool from OpenAI, including GPT-5.3 Codex Spark, and learn its CLI, IDE integration, desktop app, and web usage for secure, productive coding.
Codex requires a paid plan for full use, with a temporary free trial during the desktop release; access via ChatGPT subscriptions, other providers, or local models.
Learn to manage data privacy by adjusting the data controls to prevent your data and code from being used to train future models in ChatGPT settings.
Install Codex via npm or Homebrew on any OS, then run the Codex CLI or VS Code extension to integrate Codex in projects, using prompts, models, and files.
Explore the terminal user interface for prompt engineering, learn to queue messages with tab and escape, and use CLI and IDE integration to generate an index HTML file with Codex.
Explore the Codex desktop app for macOS, learning to create and manage projects, configure models and reasoning, insert prompts, and open in VS Code, with a multi-session workflow.
Explore the Codex desktop app to manage multiple projects, threads, and sessions with parallel agents, and customize settings, editors, git commits, and run actions; use the terminal and slash commands.
Explore Codex usage, which depends on a paid subscription with two windows: five-hour and weekly. Check remaining usage via the native app, IDE drop-down, or CLI slash status.
Configure Codex across sessions by using global or local .codex/config.toml files to set model, reasoning, personality, and project-specific settings. Also explore desktop app options and AGENTS.md for cross-session behavior.
Codex operates in a sandboxed environment with restricted file access, triggering permission popups for actions like git init and git add, while offering limited network and web search access.
Explore how to tweak Codex security settings, from dangerous config options like approval-policy to never and sandbox-mode full_access, to per-session permissions via the native app, the IDE extension, and the CLI.
Starting a new codex session creates an empty context, and you can resume older work using /resume in the CLI or from the IDE.
Guard against unwanted edits by using version control to revert to commits when undo is unreliable. Commit changes with git add and git commit, and include config.toml for shared settings.
invoke the cli with an initial prompt to start an interactive session, or run codex exec non-interactively in a git-controlled project for automation via json output.
Manage a limited context window in Codex by monitoring size, applying automatic or manual compaction, and starting a new session to preserve critical context.
Explore prompt and context engineering to craft effective prompts for Codex and other large language models, and pass only the relevant context to improve results.
Explore prompt engineering in action with spec-driven development by building a TinyNotes demo app featuring authentication, rich text notes, and public sharing, using Codex with a detailed spec.
Learn how to use Codex across CLI, IDE, and desktop app to edit a spec.md file, link to the correct file with @ and align it with Next.js server functions.
Learn how to provide precise context to Codex by referencing spec.md, including the file in prompts, and practice context engineering to improve prompt reliability.
Learn to use plan mode in codex to generate and refine task plans with follow-up questions, analyze plans, and reuse components across IDE, CLI, and native apps, including loading.tsx.
Review agent code per file, compare with the last commit in the overview pane, and use comments to revert the Inter font in the desktop app.
Create and manage Agents.md files in Codex to apply universal instructions across every session, with slash init generating defaults from the root folder so they load automatically.
Navigate nested agents.md files by placing them in subfolders like components, start codex from that folder, and understand how general and nested files apply or override in monorepos.
Leverage Codex worktrees to run tasks in separate sessions, set up a header with a logo and login and register routes, and manage branches across worktrees for clean, isolated development.
learn to add a SQLite database and create base migrations with bun, including up and down scripts for auth, through a plan mode cli that supports steering and queuing prompts.
Plan and iterate on database migrations, combining auth and notes into a single file using up-down markers, then run migrations with bun sqlite.
Connect and load up to date documentation for Better Auth with the Context7 MCP server in CodeEx, configure via config.toml, and validate the authentication flow end to end.
Discover how agent skills extend coding assistants by adding a skill.md file inside a folder to supply extra knowledge, scripts, and references for libraries and patterns.
Organize agent skills in codex using a project skills folder or a shared .agents folder, with metadata guiding lazy loading and context-aware execution for frameworks like tailwind.
Organize skills with a skill.md file and optional references in a folder, enabling lazy loading of extra documents and a script-backed SQLite explorer for safe database queries using bun.
Use the built-in skill creator to generate a code review skill, which writes a skill.md with prompts that guide the AI to perform a review and never change any code.
Learn to load skills from directories by installing recommended skills from the skills.sh repository using npx skills add, and manage them in .agents and .codex folders, with security risks.
Demonstrates adding images to prompts by integrating the Tiptap editor in a React app, and guides Codex to implement an authenticated notes UI for viewing, editing, and adding notes.
Learn to implement feedback loops that let coding agents self-validate through linting, type checks, and periodic builds, and run unit and end-to-end tests to catch and fix errors early.
Utilize the built-in code review mode to review uncommitted changes from the desktop app or CLI, enabling targeted feedback on active edits and complementing manual reviews.
Enable Codex browser access with Playwright MCP to verify changes, plan and implement updates, then test and confirm results through automated browser interactions.
Explore forking sessions in Codex to copy context between original and forked work trees, using the slash fork command, while planning two parallel plans to fix width and styling issues.
Connect GitHub to CodeX Cloud, push your main branch, and configure cloud environments with container images, preinstalled packages, environment variables, and secrets for secure deployments and pull requests.
Send tasks to codex cloud from a local project or IDE. Manage cloud tasks with list and status, then apply the changes back to your local repository.
Discover how the desktop app's automations feature lets you create prompts that run on a schedule, automatically executing Codex-driven workflows to update files like agents.md.
Codex is growing rapidly - and many developers consider it the best or amongst the best agentic engineering tools you can use these days! But unfortunately, many developers are also not using Codex very efficiently though.
If you’ve ever felt like you’re “using” Codex without really mastering it, this course is for you.
Codex – The Practical Guide is a hands-on, no-nonsense course designed to help you unlock Codex’ full potential. You’ll learn how to work with Codex efficiently, understand how it thinks, and confidently use the features that separate casual users from power users.
Yes, you can just start prompting - but to get the most out of it, you should probably go beyond basic, inefficient prompts and leverage the full suite of features Codex offers.
Starting from the fundamentals and quickly moving into advanced territory, you’ll explore essential concepts like context engineering and core commands before diving deep into powerful features such as different input modes, agent skills, MCPs, and more. Every feature is explained clearly - not just what it is, but why it matters and how to use it effectively in real projects.
You don’t need any special setup or prior experience beyond a Codex installation and subscription. By the end of the course, you’ll go from Codex novice to confident, advanced user - able to dramatically boost your productivity and build better software faster.
In this course, you’ll:
Learn the fundamentals of working with Codex the right way
Understand core commands and context engineering essentials
Master advanced, high-impact features like different input modes, agent skills, MCPs, and more
Make use of "Plan Mode" and manage configuration & permissions
Learn the WHAT, WHY, and HOW behind each concept - no black boxes
Use Codex in the CLI, IDE, native desktop app, and in the web (cloud)
Build a complete, real-world example project using Codex to see everything in action
If you want to stop guessing, stop underusing Codex, and start treating it like the powerful development tool it really is - this course will get you there.