Content modeling and ACF

When to use it

Design, debug or review WordPress content models and ACF. Use for CPT vs taxonomy vs field choices, Local JSON/PHP schema drift, missing or unformatted values, ACF validation and REST exposure, meta_query slowness and safe field renames or backfills.

Boundaries and hand-offs

Block structure and Interactivity API to wp-devkit-block-development; WPGraphQL schema and caching to wp-devkit-headless-and-wpgraphql; REST route design and permissions to wp-devkit-rest-api-development; deployment of schema changes to wp-devkit-ci-cd-and-release-engineering; broad data upgrade strategy to wp-devkit-migration-upgrade-review; XSS or authorization defects with a reachable path to wp-devkit-security-review; measured whole-site slowness to wp-devkit-performance-review.

Ask Claude

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

/wp-devkit:acf-review <target-path>

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

Ask Codex, ChatGPT or Antigravity

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

$wp-devkit-acf-and-content-modeling
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.

What it covers

The agent loads only the reference that matches the task:

• content-modeling-decisions.md : read for CPT/taxonomy/field choice and registration arguments.
• meta-queries-and-field-migrations.md : read for meta-query cost, renames, backfills.
• schema-authority-and-sync.md : read for Local JSON, PHP, database, sync and drift.
• values-validation-and-output.md : read for reads, writes, validation, REST, escaping, blocks.

How a review runs

1. Name the entities, their lifecycle needs and their consumers. Check the modeling choice against the decision table before judging field types.
2. Find the schema authority (database, Local JSON, PHP) and verify duplicates, load paths and key/name stability. Do not conclude from one file.
3. Trace each value from storage through formatting to the sink. Check return format, reference meta, escaping at output and exposure through REST/GraphQL.
4. Check every write path (admin, acf_form , REST, import, WP-CLI, update_field calls) enforces the same domain rules.
5. For queries, count participating records, inspect key/type/format, and measure or EXPLAIN before calling a query slow.
6. For a model change, map old to new, list consumers, and define backfill, dual-read, verification and rollback. Review does not run them. A finding needs a reachable path and impact. A different modeling opinion without a demonstrated cost is INFO. Insufficient evidence (no data volume, no access to the production schema) is a valid outcome: say what is missing and how to get it.

What you get back

Review: result and scope first. Each confirmed finding: severity per the contract, file:line, actor, trigger, reachable path, impact, confidence, minimal remediation, regression check. Candidates and missing evidence in a separate list. Modeling recommendations state the decision, the reason, affected consumers and migration needs.

Implementation: changed boundaries, schema authority edited, migration steps with dry-run output, commands run with exit status, checks not run, residual risk, rollback.

Where to go next

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