Code Review

My AI Development Workflow

My AI workflow is not “write one giant prompt and hope the application works.” The coding agent is only one part of the process. Most of the reliability comes from defining the project, breaking it into phases, reviewing every change, and refusing to merge until the software has been tested by a human.

I use Codex or Claude Code to plan and build, CodeRabbit for another review pass, GitHub pull requests for the final discussion, and Git as the safety net around everything. The tools can change, but the workflow stays almost the same.