Seory Posts to Markdown

FAQ

Is my content sent anywhere?

This plugin does not send your content to its developer, and it has no external service of its own. Conversion runs on your own server, and the generated ZIP is deleted right after download. Whether anything else contacts an external service depends on the content mode:

• Raw editor content (ZIP export, and get-post-markdown with mode=raw): the the_content filters are not run, so conversion contacts no external service.
• Rendered content in the ZIP export: the the_content filters run exactly as when a post is displayed. If a post contains an embeddable URL on its own line (for example a YouTube link), WordPress’s oEmbed feature may request the embed code from that provider and cache it in post meta, and your theme or other plugins hooked into the_content may make their own requests.
• Abilities (get-post-markdown with mode=rendered, and list-external-links): oEmbed is turned off while they render, so they make no oEmbed requests and do not write the oEmbed cache; embeds stay as plain URLs. Requests made by your theme or other plugins inside the_content are not blocked.

What is the difference between the two content modes?

Raw editor content converts the stored post content as-is: shortcodes stay unexpanded and block markup comments are removed. Rendered content applies the the_content filters first, so shortcodes and blocks are expanded to their final HTML before conversion.

Which user roles can use it?

• ZIP export (Tools → Seory Posts to Markdown): users with the export capability (Administrators by default).
• Abilities (for AI tools, WordPress 6.9+): any logged-in user for published posts, and Editors and Administrators for other statuses, always limited to the posts that user can read or edit. See “Which abilities are available, and who can use them?” below.

Are custom post types supported?

Yes — all public custom post types appear in the post type list.

Can I customize the front matter?

Yes. Developers can add, remove, or reorder keys with the seoryco_wpmd_frontmatter filter:

add_filter( 'seoryco_wpmd_frontmatter', function ( $pairs, $post ) {
    $pairs['subtitle'] = '"' . get_post_meta( $post->ID, 'subtitle', true ) . '"';
    return $pairs;
}, 10, 2 );

Values must be YAML-formatted; keys with an empty-string value are omitted.

Can AI tools access my content directly?

Yes, optionally. Starting with version 0.2, on WordPress 6.9+ this plugin registers read-only WordPress Abilities (wp_register_ability()) for listing posts, fetching a single post as Markdown, and listing external links. AI tools and agents (such as Claude Code or Codex) can call these through WordPress’s own REST endpoint (/wp-json/wp-abilities/v1/abilities/...) using an Application Password — the same authentication method used by the standard WordPress REST API. The request comes from the AI tool to your own site, exactly like any other REST API client; this plugin does not push your content to its developer or to any other service (see “Is my content sent anywhere?” for what rendering may contact). We recommend creating a dedicated WordPress user with limited capabilities for this purpose.

If you also install the third-party MCP Adapter plugin, these abilities can be exposed as MCP tools as well. MCP Adapter is not bundled with this plugin.

Which abilities are available, and who can use them?

All abilities are read-only and are called with GET, passing input as input[...] query parameters:

• seoryco-wpmd/list-posts — posts modified after a given time (modified_after, e.g. 2026-09-01T00:00:00Z), oldest change first
• seoryco-wpmd/list-post-ids — every post ID with its status and modified time, including trashed items, in ID order, for detecting deletions
• seoryco-wpmd/get-post-markdown — one post as Markdown with front matter, the same format as a file in the ZIP export (with mode=rendered, embeds are not fetched and stay as plain URLs)
• seoryco-wpmd/list-external-links — links to other domains with anchor text, rel, target, and surrounding text. At most 200 links are listed per post (truncated), and only the first 1 MB of each post’s rendered HTML is scanned (content_truncated, content_truncated_post_ids; developers can change the size with the seoryco_wpmd_links_max_html_bytes filter)

The list abilities are paged with cursors: pass next_cursor (list-posts, list-external-links) or next_after_id (list-post-ids) from one response as cursor / after_id in the next request, with the other input unchanged, until it is null. Posts edited or deleted while you page through do not cause others to be skipped.

“Modified time” is the effective modified time: post_modified_gmt, or, when that is 0000-00-00 00:00:00 (drafts inserted directly by some importers), post_modified converted to GMT with the site’s current UTC offset. Known limitation: if the site’s time zone switches between daylight saving and standard time while you page through list-posts or list-external-links, such posts can move by an hour and be missed in that run. A periodic full comparison with list-post-ids (for example monthly) picks them up.

Permissions are checked at two levels:

• Per request: requests limited to published posts need only a logged-in user (the read capability). Requests that include drafts, pending, scheduled, private, or trashed posts require edit_others_posts (Editors and Administrators by default). list-post-ids always requires edit_others_posts. Developers can change this request-level capability with the seoryco_wpmd_ability_permission filter.
• Per post: only publicly viewable post types are served, and a post is included only if the user can read it (read_post, for published and private posts) or edit it (edit_post, for drafts, pending, scheduled, and trashed posts). Post types with their own capabilities are checked with those capabilities. The filter above cannot widen this.

Password-protected posts are never returned as Markdown or scanned for links.