Refactor Code

Use this skill when the task is to change structure without changing intended behavior.

Core principle: Small, reversible steps over large rewrites. Separate design improvement from behavior change.

Quick Reference

HARD-GATE

NO REFACTORING WITHOUT CHARACTERIZATION TESTS FIRST.
NEVER mix behavior changes with structural refactors in the same step —
  if behavior changes are also needed, complete the structural refactor first,
  then apply behavior changes in a separate step with its own test.
ONE boundary per refactoring step — never extract two abstractions in the same step.
If a public interface changes, document the compatibility shim and its removal condition.
NEVER fabricate test output — label only actual run output as Observed output.

Core Process

1. Define stable behavior

Identify the exact inputs and outputs of the logic being refactored. Keep public interfaces stable until callers are migrated. Prefer adapters, facades, or wrappers for transitional states.

Include in your output:

• Stable behavior statement: an explicit statement of what must not change (inputs/outputs, public interfaces).
• Shim decision: name any transitional adapter/facade/wrapper and its removal condition, or state why none is needed.

2. Add characterization tests

Write this before touching any production file. No refactoring step begins until this test exists and passes on the current (un-refactored) code. If the characterization spec fails, do not continue — stop and fix the test or the behavior mismatch.

# spec/requests/orders_spec.rb  (or service/model spec — mirror the file being refactored)
# frozen_string_literal: true

RSpec.describe "Orders#create current behavior", type: :request do
  describe "POST /orders" do
    let(:valid_params) { { order: { product_id: 1, quantity: 2 } } }

    it "creates order and enqueues warehouse notification" do
      expect { post orders_path, params: valid_params }
        .to change(Order, :count).by(1)
      expect(NotifyWarehouseJob).to have_been_enqueued
    end
  end
end

Run it: bundle exec rspec spec/requests/orders_spec.rb — it must pass on the current code.

3. Choose the smallest safe slice

Good first moves include: renaming unclear methods, isolating duplicated logic behind a shared object, or wrapping external integrations before moving call sites. Add narrow seams before deleting old code paths. One boundary at a time — characterization tests first, verification after each step.

4. Execute extraction/refactor (One step at a time)

Extract, move, or rename logic. Stop and simplify if the refactor introduces more indirection than clarity.

#### Minimal Inline Example (Controller orchestration extraction) Before:

def create
  order = OrderCreator.new(params).call
  NotifyWarehouseJob.perform_later(order.id)
  redirect_to order_path(order)
end

After:

def create
  order = Orders::CreateOrder.call(params: params)
  redirect_to order_path(order)
end

Extended Resources (Progressive Disclosure)

Load these files only when their specific content is needed:

• [assets/complete_example.md](assets/complete_example.md) — A complete, step-by-step example of a high-scoring answer.md showing plan, stable behavior, characterization tests, step-by-step verification runs, and final suite verification command outputs.
• [EXAMPLES.md](EXAMPLES.md) — For more examples of extracting services and renaming in small batches.

5. Verification Protocol

Run verification after every refactoring step:

1. Run the full test suite.
2. Read the output — check exit code, count failures.
3. If tests fail: STOP, undo the step, investigate.
4. If tests pass: proceed to next step.
5. Only claim completion with evidence from the last test run — report the last line of output (e.g. "5 examples, 0 failures").

Report test run output at EACH step — not only at the end. At least two separate Observed output entries at different sequence points are required.

Evidence labelling rules: Label actual run output as Observed output only. Never use labels such as "Expected output", "Required output", "Planned output", or "Must produce 0 failures" as substitutes for actual observed run output. If you have not run the tests, you have no observed output to report.