Add React Doctor CI Buildkite job

[sejas]

Related issues

• Fixes STU-1307

Proposed Changes

• Add a new React Doctor Buildkite CI step that runs in parallel with lint, unit tests, e2e, and metrics
• New script .buildkite/commands/run-react-doctor.sh that:
  • Runs npx -y react-doctor --no-ami --yes --fail-on error to detect React code health issues
  • Parses the score from the output and fails CI if it drops below 95/100. I chose 87 as an arbitrary threshold. We can set it to any number we consider appropriate. Currently, the score is 90.
  • Posts a Buildkite annotation (error/success styled) with full diagnostics in a collapsible block
  • Skips on non-code-only changes (docs, config, etc.) via should-skip-job.sh
• Reports as a GitHub commit status check ("React Doctor")
• Add apps/studio/react-doctor.config.json to exclude Electron entry points from being omitted because they are not dead code, and omits the jsx-a11y/no-autofocus rule.
• Remove unused files and export that was not caught before to bump the score to 90.
• The job takes around 1 minute 34 seconds.

Testing Instructions

• Confirm the CI is green.
• Check the Buildkite annotation shows the score and diagnostics
• Verify the GitHub action shows the error when score < 95 or error-level issues are found. See https://buildkite.com/automattic/studio/builds/11920#019c9163-6deb-42e1-8dcd-537314da3875

Pre-merge Checklist

• Have you checked for TypeScript, React or other console errors? (Tests pass, no errors)

[katinthehatsite]

I think this always runs the latest version, right? I am wondering if we should use specific version instead? The main reason why I am thinking that is if the latest version has some breaking changes, the CI build could potentially break. But it is not a strong opinion, mostly something to consider in my opinion.

[sejas]

Yes! it uses the latest version. I had a similar concern where the CI fails because new rules are added or the score algorithm is updated, but in that case we could change the threshold if necessary. Running the latest version allows us to benefit from new best practices and bugfixes.

[katinthehatsite]

the changes look good to me. I am not sure if we need React doctor in CI so I will let others share their opinions. I left one small comment for consideration

[wpmobilebot]

📊 Performance Test Results

Comparing d59f6e0 vs trunk

site-editor

Metric	trunk	d59f6e0	Diff	Change
load	1521.00 ms	1490.00 ms	-31.00 ms	⚪ 0.0%

site-startup

Metric	trunk	d59f6e0	Diff	Change
siteCreation	7081.00 ms	7080.00 ms	-1.00 ms	⚪ 0.0%
siteStartup	3943.00 ms	3942.00 ms	-1.00 ms	⚪ 0.0%

---

Results are median values from multiple test runs.

Legend: 🟢 Improvement (faster) | 🔴 Regression (slower) | ⚪ No change (<50ms diff)

[gcsecsey]

This looks good to me, and I think this is a nice-to-have addition to our CI suite. The checks it runs are valuable on top of eg. eslint rules that are already applied locally. Also, as it's run in parallel with other jobs and usually completes in ~1.5m, I don't think it adds much overhead or potentially becomes a bottleneck on any CI run.