Address Review Comments

Address PR review comments using a test-driven approach: document the issue first, then fix it.

This playbook is tool-agnostic. Use the repository's approved git/PR workflow for your environment.

Requires: Ability to inspect PR review feedback (for example via gh or your environment's PR tooling).

Process

Step 1: Fetch Review Comments

# Get PR number for current branch
gh pr view --json number,url,reviewDecision,reviews

# Get detailed review comments
gh api repos/{owner}/{repo}/pulls/{pr_number}/comments

Step 2: Analyze Each Comment

For each review comment:

1. Understand the issue - What behavior is wrong or missing?
2. Verify validity - Is this a real bug or a misunderstanding?
3. Identify the scope - What code paths are affected?

Step 3: Write a Failing Test First (Red)

This is the critical step. Before making any code changes, document the issue as a test:

Prefer (in order):

1. Extend existing test case - If there's a related .input.tsx/.output.tsx test case, extend it
2. Create new test case - If the issue warrants a separate scenario, create new test case files
3. Add unit test - For edge cases or internal functions, add a unit test in src/__tests__/

Test case example:

// In src/__tests__/transform.test.ts
describe("issue description", () => {
  it("should handle the specific case mentioned in review", () => {
    const source = `
// Minimal reproduction of the issue
import styled from "styled-components";
const Component = styled.div\`...\`;
`;

    const result = transformWithWarnings(
      { source, path: "test.tsx" },
      { jscodeshift: j, j, stats: () => {}, report: () => {} },
      { adapter: fixtureAdapter },
    );

    // Assert the expected behavior
    expect(result.code).not.toBeNull();
    expect(result.code).toContain("expected output");
  });
});

Verify the test fails:

pnpm test:run

The test should fail, confirming the issue exists.

Step 4: Fix the Code (Green)

Now implement the fix:

1. Locate the relevant source code
2. Make the minimal change needed to fix the issue
3. Run tests to verify the fix works

pnpm test:run

Step 5: Regenerate Test Case Outputs (if applicable)

If you added a new test case or modified existing ones:

# For a specific test case
node scripts/regenerate-test-case-outputs.mts --only <case-name>

# Verify the output is correct
# Use your normal file-reading tool to inspect:
# test-cases/<case-name>.output.tsx

Step 6: Run Full Validation

pnpm check

This runs:

• Linting (oxlint + eslint)
• Type checking (tsc)
• All tests (vitest)
• Knip (dead code detection)
• Storybook build (visual verification)

Step 7: Commit and Push

git add <files>
git commit -m "fix: <description>

Addresses review comment: <brief summary>
"
git push -u origin <branch-name>

Step 8: Resolve Addressed Review Threads

After pushing the fixes, resolve the review threads that the changes addressed. Leave threads open if they were not addressed, are ambiguous, or need reviewer confirmation.

Use thread-aware PR tooling when possible so you only resolve the intended review threads, then re-fetch review thread state to verify that the addressed threads are resolved.

Step 9: Run Code Quality Refactoring

After addressing review feedback, run the refactor-code-quality playbook to:

• Remove any code duplication introduced by the fix
• Extract shared patterns if applicable
• Ensure type safety (minimize any and type assertions)

This is especially important if the fix required changes across multiple files or added new helper functions.

# Run full validation again after refactoring
pnpm check