Opleiding: ChatGPT AI with Codex
The course ChatGPT AI with Codex from SpiralTrain teaches you how to use AI as an agent that reads, changes, runs and tests code in your own repository with OpenAI Codex .
Course ChatGPT AI with Codex
The course ChatGPT AI with Codex starts with a practical introduction to the Codex CLI as an AI agent in the terminal, including the agent loop, model selection, folder trust, the composer, file mentions and shell mode. Participants learn to approve actions, use git as their undo and fork sessions, so they can work with Codex in their own repository from day one.
Context and Prompting in Codex
Next the context window is explored, covering new chats, compacting and the explore, plan, code approach. Plan mode, keeping plans in files, specific prompts, verification checks, goals, reasoning effort and web search are covered in depth. Good context and good prompts are the foundation of everything that follows in the course ChatGPT AI with Codex.
AGENTS.md, Sandbox and Approvals
This module covers project instructions with AGENTS.md files, nested instructions per folder, override files, silent truncation and memories. The sandbox then takes center stage: sandbox modes, approval policies, project trust, rules files and config profiles.