codebase-cleanup-refactor-clean

Start by identifying code smells like long methods, duplicate blocks, and tight coupling, then prioritize fixes using an ROI framework weighing business value against effort and risk. Break work into small, testable steps and validate each change with regression tests.

Common code smells include methods over 20 lines, classes over 200 lines, duplicate code blocks, dead code, magic numbers, long parameter lists, and feature envy. Each smell has a standard remedy such as method extraction, parameter objects, or moving methods to the class they depend on.

Apply SOLID by giving each class one responsibility, using strategy interfaces for extensibility, respecting substitutability in inheritance, segregating fat interfaces, and depending on abstractions via dependency injection. The playbook includes concrete before/after examples for each principle.

Avoid refactoring when you only need a tiny targeted fix, when policy or deadlines block changes, or when the request is documentation-only. Large rewrites without agreed scope and test coverage also carry high regression risk.

Warning signs include cyclomatic complexity above 10, methods over 20 lines, classes over 200 lines, test coverage below 60 percent, and duplication above 3 percent. Critical thresholds like complexity over 15 demand immediate decomposition.