Loading TutorKit...
Using Codex, Claude Code, and Cursor
How terminal and editor coding agents like Claude Code, Codex, and Cursor inspect a repository, plan a change, edit files, run commands, and get verified — and the practical discipline of directing, sandboxing, and reviewing them without surrendering engineering judgment.
Questions in this set
- What is an AI coding agent?
- How do Codex, Claude Code, and Cursor differ?
- What can a coding agent actually see?
- How should you start a coding task?
- Why should the agent inspect before editing?
- How should repository instructions be written?
- How do you control the size of a change?
- When should the agent run commands?
- How do approvals and sandboxing protect the project?
- How should you review an agent's diff?
- How do you recover when the agent goes in the wrong direction?
- How do you compare coding agents on your codebase?
- What should remain the developer's responsibility?
TutorWant me to explain it differently?
AI concepts can be dense. Tell me what's confusing and I'll find a new analogy.