Skip to the guide
All guides

Admin interfaces

Find why a WordPress settings page saves wrongly, who can really change its data, and how its states behave.

app WordPress DevKit 2.0.3checked reading 4 minutes

On this page

When to use it

Use this skill for WordPress admin settings pages, list tables, custom save handlers and editor-side tools. It fits problems such as a save that says success but changes nothing, users with the wrong role or capability, assets that load on every admin screen, notices, and loading or error states that people with assistive technology cannot use.

It does not design REST routes.

This skill owns the admin screen, its save boundary and its interaction states. Other topics go to a neighbor.

  • REST route design and authorization: wp-devkit-rest-api-development
  • Plugin bootstrap and storage: wp-devkit-plugin-development
  • Blocks in the block editor: wp-devkit-block-development
  • Formal WCAG evidence: wp-devkit-wcag-review
  • Accessibility defects in code: wp-devkit-accessibility-review
  • Visual design direction: wp-devkit-ui-design

Run a review with Claude

With the plugin loaded, start a read-only review of a bounded target:

Text
/wp-devkit:admin-review <target-path>

The short form /wp-devkit:admin requests a quick scan.

Run a review with Codex, ChatGPT or Antigravity

Name the installed skill and keep the request small. Replace the target with your files or journey.

Text
$wp-devkit-admin-ui-development
Review the target without editing files.
Read the project instructions and the skill's engineering contract first.
Separate confirmed findings from candidates and state the evidence and checks you could not run.

The engineering contract is the set of rules every DevKit skill follows, such as staying read-only during a review and sorting findings by how well they are proven.

Reference files the skill loads

The skill reads the workbook first, then only the topic files your task needs.

  • admin-ui-development-workbook.md helps you go from a symptom to a cause. It has worked decisions, common false alarms and verification steps.
  • settings-and-save-flows.md covers the Settings API, sanitizing versus validating, the capability check in options.php, admin-post and AJAX handlers, REST-backed settings and secrets.
  • screens-assets-and-lists.md covers menus and access, loading assets only on the right screen, enqueue arguments, translations, React components, WP_List_Table, notices and the admin changes in WordPress 7.0.
  • admin-accessibility-and-verification.md covers labels, status announcements, focus, the WCAG 2.2 criteria that apply, and checking with Playwright, axe and a role matrix.

How a review runs

  1. The skill follows screen registration and each write handler separately. A hidden menu item or a blocked direct URL does not protect an admin-post, AJAX or REST write.
  2. It traces the save path: option group, registered setting, posted field names, sanitizer, validation, nonce and capability. It confirms that earlier valid data survives a failed submission.
  3. It traces assets to the exact screen (using the hook suffix) and to their dependencies. It prefers the WordPress components that are already installed, and does not add a second UI framework for a small change.
  4. It assesses labels, errors, pending, success and empty states, focus and announcements. It gathers keyboard evidence and describes how to check a double submit without making live writes.
  5. It reports the screen and save boundary, the confirmed behavior, and a scoped correction with permission and interaction checks. Options, metadata and assets stay unchanged during a review.
  6. It sorts findings by how well they are proven. A pattern match stays a candidate, and the result is "insufficient evidence" when the save path, the role set or the version is unknown.

What you get back

A review starts with the result and the scope reviewed. Each confirmed finding lists the file and line, who triggers it, the reachable path, the impact, the confidence, the severity, the smallest fix and a regression check. Candidates and missing evidence are listed apart.

An implementation report lists the changed boundaries, the role and interaction checks with results and exit status, the checks not run and the remaining risk.

Note

Reviews are read-only. A clean review covers the inspected scope and is not a certification. Ask for implementation as a separate, bounded task with a regression check.

Where to go next

Use review to fix for a bounded implementation and Doctor and reports for environment and evidence tools.