Learn agentic coding using Codex and AI coding agents.
2-hour course teaches Codex and AI coding agents with prompting, AGENTS.md, Plan Mode, testing, secrets, permissions, config.toml, and external tools
Key Takeaways
- A 2-hour freeCodeCamp.org course by Tom Chant teaches Codex coding agents for experienced developers and non-coders who want to build software with AI assistance.
- The course focuses on a practical workflow: giving the agent proper context, planning tasks, controlling permissions, and verifying every change.
- Core topics include AGENTS.md, Plan Mode, skills, testing agent changes, Git safety, secret handling, config.toml, VS Code Codex, Model Context Protocol, and shadcn.
- It is best suited for users who want a structured way to add coding agents to real projects, not a general programming introduction or a short prompt list.
A 2-hour watch course has just been published on the freeCodeCamp.org YouTube channel. It is taught by Tom Chant, Head of Education at Scrimba. The course is for experienced developers who want to add a Codex coding agent to their workflow and for non-coders who want to build software with AI help. It aims to teach reliable results by giving the agent the right context, planning the work, controlling its permissions, and verifying every change.
What the course covers
Coding agents such as Codex can read a codebase, edit multiple files, and run commands and tests while working through larger tasks. The course begins with Prompting & Project Setup. It covers how to write effective prompts, how to comment prompts for a coding agent, how to initialize a project, and how to use multiple AGENTS.md files. AGENTS.md is described as a file that holds persistent instructions for a project.
The next section is Plan Mode & Skills. Plan Mode is used to break larger development tasks into manageable steps, while skills extend what a coding agent can do. The course then covers Testing Agent Changes. It explains how to verify an agent's changes with characterization tests and test-driven development, and which testing pitfalls to watch for.
Git & Secrets is the next topic. It teaches how to use Git safely alongside a coding agent and how to keep API keys, credentials, and other secrets safe. The topic is placed alongside testing and permissions, which gives the course a safety-focused structure.
The course also covers Context, Permissions & config.toml. config.toml holds the agent's configuration. It explains how to use the VS Code Codex extension, manage context, control what an agent can access and change, and configure the agent with config.toml. These settings matter because the agent does not only suggest code; it can interact with a project while working through tasks.
Finally, the course covers Model Context Protocol & shadcn, showing how to connect agents to external tools. The module follows context and permissions in the course outline, so tool connections appear after the agent's access and configuration have been explained.
Learners work through realistic tasks, including challenges on skills, characterization tests, and config.toml. The full course can be watched on the freeCodeCamp.org YouTube channel as a 2-hour watch, and it can also be studied interactively on Scrimba.
Why the workflow matters
Taken together, the course presents a workflow rather than a single trick. Clear prompts and project instructions give the agent direction. Plan Mode helps when tasks are too large for one step. Skills extend capabilities. Characterization tests and test-driven development provide evidence that changes behave as expected. Safe Git habits and secret handling reduce the risk of accidental exposure or unwanted changes. The VS Code Codex extension and config.toml address context and permissions, which are especially important when an agent can access files and run commands. The Model Context Protocol and shadcn cover connections to external tools.
For developers, the practical value is process. A coding agent can move quickly across files and commands, so the course treats context, permissions, planning, and verification as parts of the same job. Instead of asking only for a final answer, the workflow gives the agent a defined project, a sequence of steps, a way to test changes, and boundaries for what it may access.
For non-coders, the course provides a structured path through the same components. The source describes the audience as people who want to build software with AI help, and it teaches project files, configuration, tests, secrets, and external tools as learnable steps. The source does not provide benchmark numbers or performance claims; it describes tasks, tools, and verification steps.
Who should use it / who should skip it
Use this course if you are an experienced developer adding AI to your workflow or a non-coder who wants to build software with AI help. It is also useful if you already know the idea of coding agents but need a structured explanation of AGENTS.md, config.toml, Plan Mode, skills, tests, Git safety, secrets, VS Code Codex extension, Model Context Protocol, and shadcn.
Skip it if you are looking for a general introduction to programming rather than a course about using coding agents. The source describes the audience as experienced developers and non-coders building software with AI help, not as a beginner programming curriculum. If you only want a short list of prompts, the course may be too broad, because it also covers project setup, testing, permissions, secrets, and external tool connections.
Meta description: A 2-hour freeCodeCamp.org course by Tom Chant of Scrimba teaches Codex coding agents, AGENTS.md, config.toml, Plan Mode, tests, and Model Context Protocol.
Tags: Codex, coding agents, freeCodeCamp, Scrimba, Model Context Protocol
Featured image: Abstract, friendly illustration of code files, a test checklist, and a configuration panel, with no real people, no logos, and no text.
Comments (0)