Safe Legacy Editing Discipline

Apply 4 editing disciplines when modifying legacy code: Hyperaware Editing, Single-Goal Editing, Preserve Signatures, Lean on the Compiler. Use whenever a developer is about to edit untested code, refactor without a safety net, or is caught in avoidance anti-patterns (Edit-and-Pray, Minimization Freeze, Legacy Code Dilemma paralysis). Activates for 'how do I refactor safely', 'edit legacy code without breaking it', 'preserve behavior during refactor', 'lean on the compiler', 'preserve signatures', 'single-goal editing', 'make safe changes', 'edit and pray', 'cover and modify', 'careful editing', 'refactor without tests', 'break dependencies safely', 'avoid regressions in legacy code'.

View on ClawhHubView on GitHub

Install

Claude CodeClawhHubnpxCursorCodexGemini CLI

1. Add marketplace

› /plugin marketplace add bookforge-ai/bookforge-skills

2. Install plugin

› /plugin install working-effectively-with-legacy-code@bookforge-skills

3. Use the skill

› /safe-legacy-editing-discipline

CC-BY-SA · Open sourceGitHub

Source Book

Working Effectively with Legacy Code

Michael C. Feathers

More from Working Effectively with Legacy Code

Legacy Codehybrid

Big Class Responsibility Extraction

Identify and extract responsibilities from an oversized class using Feathers' 7 heuristics + feature sketches. Use whenever a developer faces 'this class is too big', 'god class', 'monster class', '500-line class', 'class with 50+ methods', 'class with too many responsibilities', 'SRP violation', 'hard to test because class does too much', 'team keeps editing same class', 'merge conflicts on one class'. Activates for 'single responsibility', 'extract class', 'class decomposition', 'responsibility split', 'interface segregation', 'class bloat', 'swamp class', 'god object', 'feature sketch', 'class too big'.

+67%

Legacy Codehybrid

Change Effect Analysis

Trace the blast radius of a legacy code change and produce a test placement plan with pinch points. Use whenever a developer needs to decide WHERE to write tests for a pending change — 'what should I test?', 'where to test?', 'how far do the effects propagate?', 'what else could this break?', 'how to find test points', 'pinch point', 'effect sketch', 'impact analysis', 'blast radius', 'interception point', 'high-leverage test'. Triggers for 'I need to make many changes', 'many classes affected', 'cluster of related changes', 'cross-class refactoring'.

+71%

Legacy Codefull

Characterization Test Writing

Write tests that pin down the actual current behavior of untested legacy code as a safety net for change. Use whenever a developer needs to create a regression safety net before modifying code — 'I don't know what tests to write', 'what should I test in legacy code', 'how do I write tests for code I didn't write', 'tests to preserve behavior', 'golden tests', 'snapshot tests for old code', 'characterization test', 'cover before modify'. Activates for 'legacy code testing', 'untested code', 'write tests before changing', 'regression safety net', 'behavior-preservation tests', 'pin down behavior'.

+72%

Legacy Codefull

Dependency Breaking Technique Executor

Select and execute the right dependency-breaking technique from Michael Feathers' catalog of 24 named techniques (Part III of Working Effectively with Legacy Code) for a specific testability obstacle. Use when a class or method cannot be placed under test due to a hard-coded dependency, and you need concrete step-by-step mechanics for breaking it safely — without existing tests protecting you. Activates for 'can't test this class', 'constructor dependency', 'global variable blocking tests', 'static call in method', 'inject a fake', 'break dependency for testing', 'extract interface legacy', 'subclass and override', 'parameterize constructor', 'encapsulate global', 'dependency injection without framework', 'extract and override', 'how to fake this in test', 'legacy code testability refactoring'.

+63%