What problem does it solve? Structural code changes often cause silent behavior regressions, scope creep, and untraceable bugs when moves, renames, and logic edits are mixed together. This Skill enforces a disciplined refactor process that locks in existing behavior with tests before any change and keeps every step independently verifiable and reversible. ## Core Features & Use Cases - Green baseline and characterization tests: Record pass/fail counts and coverage before touching code, and pin current behavior with tests committed first. - Atomic step sequencing: Split renames, file moves, signature changes, and logic edits into separate commits, re-running scoped tests after each one. - Call-site and hidden-consumer mapping: Find every caller with git grep, then check string literals, dynamic imports, mocks, fixtures, docs, and CI files that compilers miss. - Risk-tiered rigor: Calibrate the safety net from low-risk local renames up to critical ORM or auth migrations requiring feature flags and staged rollouts. - Use Case: When renaming an exported pricing function used across a package, map all call sites, commit the file move and symbol rename separately, verify each caller, and prove behavior is preserved with an identical before/after test result. ## Quick Start Ask the AI to refactor the pricing module by renaming calcPrice to priceCalculator while preserving behavior, following the refactor-safety procedure with a recorded test baseline and atomic commits.
