wordpress-content-sync

Sync WordPress posts, pages, ACF fields, comments and media between two sites through the Content Sync, Media Sync, Import, Export plugin Abilities API (rsl-ie/*). Use when the user asks to push, pull, copy or update content between a staging/production or any two connected WordPress sites without export files.

WordPress Content Sync (rsl-ie abilities)

Requires the free plugin "Content Sync, Media Sync, Import, Export with AI Agents Support" (import-export-by-rockstarlab) on BOTH sites, WordPress 6.9+ Abilities API, and an administrator application password or the MCP Adapter. Products, CPTs, WooCommerce: see woocommerce-sync-pro.

Call abilities through the MCP Adapter, or REST: POST /wp-json/wp-abilities/v1/abilities/rsl-ie/<name>/run with {"input":{...}}. Read-only abilities use GET with ?input[key]=value.

Workflow

  1. rsl-ie/get-capabilities (check plugin version, abilities available).
  2. rsl-ie/connection-list → pick site_id of the target. If none: connection-create needs the other site's incoming key (site-key-status / site-key-reveal on that site; reveal only on explicit user request, never print the key). Verify with connection-test.
  3. Choose records: sync-browse (remote), sync-search-local, or ids given by the user. post_type is post or page in the free plugin.
  4. Inspect: sync-local-info (post_type, post_ids ≤50) shows what exists and original ids.
  5. Optional mapping to update existing content instead of creating copies: sync-auto-map (exact normalized title match) and pass the result as post_mapping. Show the user what will be updated before running.
  6. Start: sync-push-start (local → remote) or sync-pull-start (remote → local). Required: site_id, post_type, post_ids (1–50), request_id (unique string; reuse the SAME value only when retrying the same request, it deduplicates).
    • Posts with ACF image/gallery fields: set acf_media_references: true.
    • Hierarchies: pass hierarchy so parents map correctly.
  7. Poll sync-job-status (job_id) until done. Runs on WP-Cron, so a long wait on a low-traffic site is normal; do not start a second job. Details afterwards: sync-result (source/target/media id maps, works after HTTP timeouts).
  8. Verify on the target: sync-browse or open the returned target ids; compare titles, status, featured image, ACF values on a sample.

Comments: sync-comments-browse, then sync-comments-push-start / sync-comments-pull-start (≤50 ids). Media only: sync-media-browse, sync-media-push-start / sync-media-pull-start.

Rules

  • These calls write to the destination site. Confirm direction (which site is overwritten) and the list of records with the user before starting, unless they already stated both.
  • ≤50 records per job; split larger sets into batches and run them one after another.
  • Never retry blindly after an error: check sync-job-status/sync-result first, since the job may have finished.
  • sync-job-pause, sync-job-resume, sync-job-cancel control running jobs. Completed changes are not rolled back by cancel.
  • Synchronous sync-push / sync-pull exist but may hit HTTP timeouts; prefer the -start versions.
  • Take a backup of the destination before overwriting existing content.