refactor

To refactor code safely, you need a disciplined workflow of small, atomic changes with passing tests verified after each step. This ensures module-wide refactors maintain stable behavior through explicit rollbacks and traceable commits.

The best way to plan a code refactor is stepwise: perform pre-flight checks for a clean working tree, write a test-backed plan, and apply one atomic change at a time. This keeps changes isolated and traceable.

Yes, you need passing tests before refactoring. Pre-flight checks verify your working tree is clean and tests are passing, establishing a reliable baseline to detect any unintended behavior changes during the refactor.

You verify code quality during refactoring by running tests after each atomic change. Guardrails for test coverage, naming conventions, and explicit rollbacks ensure changes are traceable and behavior remains stable.

Yes, atomic commits enable explicit rollbacks for a failed code refactor. By committing small, isolated changes with version control, you can safely revert to a known stable state if tests fail.

You should avoid large single-step refactors because they risk changing behavior. Instead, break the module-wide refactor into small, atomic changes, verifying tests and committing after each step to ensure stability.