Programming Study日本語

Disclosure: We earn a commission from Udemy when you buy a course through links on this page.

Claude Code: AI Refactoring for Legacy Code

Claude Code: AI Refactoring for Legacy Code

Udemy · English

About this course

Language
English
Category
Provider
Udemy
Data updated

Build a reviewed legacy-code change with tests and rollback evidenceWork through one end-to-end AI refactoring workflow with Claude Code. In the Northstar Python lab, you will record a baseline, check existing behaviour, review one bounded patch and rehearse rollback. The result is a tested modernization slice with evidence another developer can inspect and reproduce. Five focused video lessons, with 36 minutes of video, take you through the same practical workflow. Set aside additional time for hands-on practice in the lab, running the checks and reviewing your results. The Northstar settlement lab starts with a one-penny difference between a legacy batch and its proposed replacement. You will reproduce the mismatch and record the starting state before allowing Claude Code to edit code. You will trace the rounding rule and capture what the existing code does in a characterisation record. You will also distinguish observed behaviour from an approved business rule: matching old output alone does not establish what the system should do. Next, you will restrict the files Claude Code can change and review a bounded correction in a Git worktree. The legacy implementation and reference output (golden master) remain protected. You will inspect the complete Git diff and compare the old and replacement outputs with differential tests. The practical deliverables are a recorded baseline, a scoped characterisation record, a reviewed patch and a small evidence bundle. You will rehearse patch-based rollback so another developer can inspect the change and reproduce your checks. These checks help you assess a change; they do not guarantee that every defect has been found. The companion lab uses fictional records, Python 3.11 or later, GNU Make and Git. The core exercise has no third-party Python dependencies and makes no network calls. Claude Code is required for the guided analysis and implementation session.

Courses in the same category