Map entry points before a risky change
Claude Code handles this as a draft: Scripts, main files, and the test command in the README. Do not push.
- Steps
- 4
- Cost
- low
- Time
- Half an hour
Unverified · code.claude.com · 16 Oct 2026
Loading the catalog…
Loading the catalog…
Claude Code reads a single module, writes characterization tests, then changes structure only where those tests pass.
Use the current quickstart at code.claude.com. Run it from the repository root so the agent sees the real test script.
A whole service is too large for a first refactor. Pick a module with a test runner already configured.
If it starts editing production code before tests exist, stop the session and restate step 2.
Passing tests are not a review. Look for behavior hidden in snapshots or deleted assertions.
Claude Code can draft the branch. You merge.
These are editorial labels so you can compare playbooks. They are not measurements.
No. Status is unverified. Paste the prompt, keep the version that worked, and do not treat the time or cost lines as measurements.
No. Constrain the directory. A wide prompt produces a diff nobody can review in one sitting.
Official documentation URL, checked when this seed was written. The prompt and steps are an original AI Playbook Gems playbook, not a quotation of a customer story, and have not been re-run.
Machine-readable copy: JSON
Claude Code handles this as a draft: Scripts, main files, and the test command in the README. Do not push.
Unverified · code.claude.com · 16 Oct 2026
Claude Code handles this as a draft: Which commands are missing a script entry. Do not rewrite the README until I agree.
Unverified · code.claude.com · 14 Oct 2026
Claude Code summarizes breaking changes for a single dependency, applies the upgrade, and runs the existing tests. Major bumps stay behind a question.
Unverified · code.claude.com · 12 Oct 2026