ai-slop-cleaner

AI code cleanup requires a regression-tests-first workflow to preserve behavior while reducing slop. You provide a cleanup plan and changed-files list, then execute staged passes for specific smells before running regression tests and lint checks.

A regression-tests-first workflow means establishing tests that lock current behavior before removing AI code smells. This ensures that staged refactoring passes for duplication, dead code, or needless abstraction do not introduce functional regressions.

You remove code smells from AI edits by bounding the cleanup to a changed-files list or Ralph workflow to limit scope. This allows targeted passes for duplication, dead code, needless abstraction, and boundary violations one smell at a time.

Yes, AI slop cleanup requires a defined plan before starting. You must provide an initial cleanup plan to bound the scope, then start the first pass on targeted files to execute smell removal, regression tests, and lint checks systematically.

Code smells targeted during AI refactoring include Duplication, Dead code, Needless abstraction, and Boundary violations. The workflow executes staged passes to resolve these issues one smell category at a time, maintaining structured and disciplined cleanup.

You document changes and remaining risks after refactoring AI code by generating an evidence-dense report. This report records applied changes, executed regression tests, and remaining risks discovered during the staged smell removal passes.