Software Engineering

AI-supported planning and testing

Before writing code, AI is prompted to think through and plan an approach, and a separate agent can be used to review or test the generated output.

Why the human is still essential here

The engineer still decides the approach, judges whether the plan makes sense, and reviews test results and code quality before accepting changes.

How people use this

Ticket-to-implementation plan

AI turns a feature request or bug ticket into a step-by-step implementation plan, identifying files to touch, edge cases, and dependencies before coding starts.

Claude / ChatGPT

Unit test scaffold generation

AI proposes focused unit tests, mocks, and fixtures for the generated code so the engineer can validate behavior and fill in any missing cases.

GitHub Copilot / Cursor

Independent pull request review

A separate AI reviewer inspects the diff for logic issues, regressions, and missing tests before merge, giving the engineer a second pass on the change.

CodeRabbit / GitHub Copilot

Related Prompts (4)