02-llm-editing-workflow/04-how-chatgpt-should-refactor-an-app.md

How ChatGPT should refactor an app

Refactoring means improving the structure of the code **without changing its behavior**.

In Vroq apps, refactoring should move code toward the house style while keeping the system stable and easy to debug.

When refactoring is appropriate

Refactor when:

a file has become too large
responsibilities are mixed in one file
logic is duplicated
naming is unclear
API or state logic is leaking into UI files

Refactoring should improve clarity and separation of concerns.

Preferred refactoring workflow

1. understand the existing behavior 2. identify a clear responsibility boundary 3. extract the code into a new file or section 4. keep the original file as the orchestrator 5. verify that behavior is unchanged

Typical refactoring moves

Common safe refactors include:

extracting a section from a large feature file
moving API calls into a dedicated API client file
splitting reducer logic into clearer cases
renaming unclear variables or functions
moving reusable logic into shared helpers

Avoid risky refactors

Avoid large architectural rewrites unless the structure is clearly broken.

Large rewrites often introduce regressions and make debugging harder.

Prefer **incremental refactoring**.

Preserve behavior

During refactoring:

do not change state shape unless necessary
do not rename actions without updating all usage
keep API behavior consistent

Behavioral changes should happen in separate commits or edits.

Final rule

Refactor in small, safe steps that make the code easier to understand and easier to change later.