OpenAI Codex — Real-World Software Engineering
Lesson 3: Feature Work, Bug Fixes, Tests, and Code Review
Lesson Objectives
By the end of this lesson, students should be able to:
- Write a testable engineering task
- Require inspection and plan before modification
- Use tests, linting, static analysis, and targeted manual checks
- Review diffs for correctness, scope, security, and maintainability
Lesson Content
A strong Codex task includes the observed problem, expected behavior, reproduction steps, affected users, constraints, likely files if known, tests to add, and definition of done. Ask Codex to inspect the repository and summarize its plan before editing when the task is complex or risky.
The implementation loop is: reproduce, identify cause, propose plan, change the smallest reasonable surface, add or update tests, run validation, inspect failures, revise, and present a concise change summary. Do not accept “tests pass†without seeing which tests ran and their result.
Review the diff, not just the summary. Look for unrelated changes, disabled safeguards, hard-coded values, missing error handling, insecure data flow, migration risk, accessibility regressions, and changes to generated or vendor files. Request a second review focused on security and edge cases, but do not treat self-review as independent assurance.
Enroll to continue this lesson.
The preview above shows the lesson objectives and opening lesson content. Enroll to view the full lesson, complete the practice work, and take the lesson quiz.