ai-slop-cleaner

To remove AI-generated code slop without changing behavior, use a regression-safe cleanup workflow that prioritizes deletion, deduplication, and small reversible diffs. It locks intended behavior with narrow regression tests before applying edits.

The best way to clean up over-abstracted and repetitive AI code is to apply smell-focused passes that simplify logic and remove dead code, followed by running quality gates like lint, typecheck, and unit tests to verify integrity.

Regression testing during AI code deduplication involves adding or running narrow tests to lock intended behavior before editing. This protect-first strategy ensures that boundary tightening and logic deduplication do not alter functionality.

Yes, you can use a reviewer-only mode for dead code removal by enabling the --review flag. This enforces separation of duties so higher-impact cleanup and refactoring changes are not self-approved in a single pass.

When tightening boundaries in weakly tested AI code, limitations include the risk of breaking intended behavior if regression tests are absent. The workflow requires writing a cleanup plan and running available quality gates after each pass to mitigate this.

You do not need existing unit tests to start a code cleanup workflow. The verification-first approach adds or runs narrow regression tests to lock intended behavior before executing any deletion or refactoring edits.