audit-blocks-deep

name	audit-blocks-deep
description	Deep AI audit of all blocks — runs structural checks first, then uses AI to review content quality, descriptions, grammar, admin render justification, wrapper usage, and empty-state messages.

Read [@/knowledge/acf-blocks/conventions.md](/knowledge/acf-blocks/conventions.md) and [@/knowledge/acf-blocks/fields.md](/knowledge/acf-blocks/fields.md), then:

Step 1 — Structural check

node $(claude plugin path wp-blocks-dev)/scripts/audit-blocks.js

If there are ✗ issues, tell the user to fix them before proceeding. Warnings are fine to continue.

Step 2 — AI content review

For each block read block.json, fields.json, render.php, render-admin.php (if exists) and check:

• Title and description — clear, human-readable, no placeholders, no grammar errors
• ACF field labels — meaningful, no defaults like "Text" or "Field Label", no typos. Never suggest changing name or key.
• Wrapper — render.php must use get_block_wrapper_attributes(), never build wrapper attributes manually
• Admin render justification — render-admin.php only for interactive/hidden-content blocks; flag static blocks that have it unnecessarily, and interactive blocks that are missing it
• Empty states — render-admin.php must show a placeholder when content is empty; render.php must render nothing on the frontend when content is missing
• render.php cleanliness — no commented-out code, no leftover scaffold comments

Output format

{block-name}
  Structural:   ✔ passed
  Title:        ✔ clear
  Description:  ✗ "Sample description" — generic, update it
  Fields:       ⚠ field "text_1" has a generic label
  Wrapper:      ✔ correct
  Admin render: ✔ justified
  Empty states: ⚠ missing placeholder in render-admin.php

End with a summary count, then ask: "Fix all issues automatically? (or specify which blocks/rules to fix)" — only proceed after confirmation.