ePublishing Ask My Brand Feed

By askmybrand

20+ active installationsVersion 1.6.0

Provides a secure, full-content RSS feed for Ask My Brand integration.

Description

This plugin creates a secure RSS feed endpoint that delivers full post and page content for integration with ePublishing’s Ask My Brand AI-powered search.

Features:

  • Token-based authentication to protect your content
  • Full content (not just excerpts)
  • Posts and pages in a single feed, each item labelled with its content type
  • Pagination support for large archives
  • Category and content type filtering, plus a per-site category exclusion list
  • Last modified dates for efficient sync

Every item carries an <askmybrand:content_type> element — Article for posts, Page for pages — so the indexer can tell them apart. Pages can be excluded entirely from Settings → Ask My Brand.

External Services

This plugin is designed to integrate with Ask My Brand, a third-party AI-powered search service provided by ePublishing.

What the service is and what it’s used for:
Ask My Brand is an AI-powered content search and discovery service that indexes your WordPress content to provide intelligent search capabilities for your audience. The plugin creates a secure RSS feed that the Ask My Brand service accesses to retrieve and index your published content.

What data is sent and when:
When Ask My Brand’s service accesses your feed (using the token-authenticated URL you provide), the following data is transmitted:
– Published post and page titles
– Published post and page content (full text, with <script> and <style> blocks removed)
– Post and page URLs
– Publication dates and last modified dates
– The content type of each item (Article or Page)
– The URL, dimensions and caption (or alt text) of each item’s featured image
– Post categories
– Post and page authors

Site code can extend the feed to further content types (for example WooCommerce products) using the epubamb_content_types filter described under “For developers”; doing so adds those items’ titles, content and URLs to what is transmitted. No content type beyond posts and pages is included unless the site explicitly adds it.

Categories can be excluded from the feed entirely in Settings → Ask My Brand; posts in an excluded category, or in any of its child categories, are never transmitted.

This data is only sent when Ask My Brand’s servers request the feed URL. No data is automatically transmitted without the feed being accessed. The feed only includes published posts and pages, excludes password-protected content, excludes anything marked noindex in Yoast SEO or Rank Math or ticked “Hide from Ask My Brand” on the edit screen, excludes items with neither content nor an excerpt, and respects WordPress content visibility settings. Pages can be excluded from the feed entirely in Settings → Ask My Brand.

Service provider information:
– Service: Ask My Brand by ePublishing
– Terms of Service: https://askmybrand.ai/terms-of-service
– Privacy Policy: https://askmybrand.ai/privacy-policy
– Service Website: https://askmybrand.ai/

By using this plugin and providing the feed URL to Ask My Brand, you consent to share your published content with the Ask My Brand service for indexing and search purposes.

Usage

Once activated, your feed is available at:

/feed/ask-my-brand/?token=YOUR_TOKEN

With Plain permalinks it is /?feed=ask-my-brand&token=YOUR_TOKEN instead; Settings → Ask My Brand shows the exact URL for your site.

Parameters:

  • paged — Page number (default: 1)
  • per_page — Items per page, max 100 (default: 50)
  • category — Category slug or comma-separated slugs (default: all)
  • content_type — Article, Page, or any label a developer has added, comma-separated (default: every type enabled in settings)

Examples:

  • /feed/ask-my-brand/?token=xxx&paged=2&per_page=25
  • /feed/ask-my-brand/?token=xxx&category=news
  • /feed/ask-my-brand/?token=xxx&content_type=Page
  • /feed/ask-my-brand/?token=xxx&category=news,blog&per_page=100

Notes:

  • Pages carry no categories, so filtering by category returns posts only.
  • Categories excluded in Settings → Ask My Brand are removed before any category parameter is applied, and are reflected in totalPosts / totalPages. Excluding a parent category also excludes its children.
  • If pages are disabled in Settings → Ask My Brand, content_type=Page returns an empty feed rather than falling back to posts.
  • An unrecognised content_type value is ignored, and the feed returns every enabled type.
  • Requesting a page number past the end of the results returns an empty feed with a 404 status; treat that as end-of-stream.

Feed format

The feed is standard RSS 2.0, so any RSS parser reads it. The additions live in the askmybrand namespace, declared once on the root element:

<rss version="2.0" xmlns:askmybrand="https://askmybrand.ai/feed" xmlns:media="http://search.yahoo.com/mrss/" ...>

Featured images use the standard Media RSS (media) namespace.

Channel level. One <askmybrand:pagination> block describes the whole result set, so a crawler can size the job before walking it:

<askmybrand:pagination>
  <askmybrand:currentPage>1</askmybrand:currentPage>
  <askmybrand:perPage>50</askmybrand:perPage>
  <askmybrand:totalPosts>130</askmybrand:totalPosts>
  <askmybrand:totalPages>3</askmybrand:totalPages>
  <askmybrand:contentTypes>Article,Page</askmybrand:contentTypes>
</askmybrand:pagination>
  • totalPosts and totalPages describe the filtered set, not the whole site — request ?category=news and they count only that category.
  • contentTypes lists the types this particular request can actually return. It is not a list of what the site has: under a category filter it drops to Article, because pages and most custom types carry no categories.
  • categories is present only when you passed category, echoing back the slugs that were applied. It reports the slugs actually in force, so a site whose epubamb_feed_query_args filter replaced them will not see it.

Item level. Each <item> carries the standard RSS fields plus these:

<item>
  <title>Example post</title>
  <link>https://example.com/example-post/</link>
  <pubDate>Mon, 15 Sep 2025 12:00:00 +0000</pubDate>
  <dc:creator><![CDATA[admin]]></dc:creator>
  <category><![CDATA[News]]></category>
  <content:encoded><![CDATA[<p>The full post content.</p>]]></content:encoded>
  <lastModified>Tue, 16 Sep 2025 09:30:00 +0000</lastModified>
  <askmybrand:content_type>Article</askmybrand:content_type>
  <media:thumbnail url="https://example.com/wp-content/uploads/example-300x200.jpg" width="300" height="200" />
  <media:description>A caption for the image</media:description>
</item>
  • <content:encoded> is the full content, never an excerpt, regardless of Settings → Reading — unless the item has no content, in which case WordPress sends its excerpt. Content after a <!--nextpage--> break is included; <script> and <style> blocks are removed.
  • <lastModified> is when the item last changed, in GMT: the later of its last edit and its publication, so a scheduled post counts as changed when it goes live. Use it for incremental sync. <pubDate> remains the publication time.
  • <askmybrand:content_type> is Article for posts, Page for pages, or whatever label the site has registered for a custom type. It is omitted, rather than guessed at, for a type the feed has no label for.
  • <media:thumbnail> is the item’s featured image at WordPress’s medium size, and <media:description> its caption, falling back to its alt text and then its title, as plain text. Both are omitted when the item has no featured image; <media:description> is also omitted when all three are empty.

Crawling. Items are ordered by <lastModified>, newest first, so an incremental sync can stop at the first item whose <lastModified> is older than the moment its last run started — an old page edited yesterday is at the top, not buried by its publication date. Measure from the start, not the end: an item edited mid-crawl jumps to page 1, which that run has already fetched, so only the next run sees it. Walk paged from 1 until currentPage reaches totalPages. Requesting a page past the end returns an empty but well-formed feed with a 404 status and no pagination block — treat that as end-of-stream rather than an error, and never as the site having no content.

Excluded content. Drafts, pending, private and password-protected items never appear, and neither do attachments. The feed carries only published, publicly visible content. Items marked noindex in Yoast SEO or Rank Math are excluded too — per item, or through the post type’s default, following the same rules those plugins use for the page’s robots tag — as is anything ticked “Hide from Ask My Brand” in the edit screen’s Ask My Brand box, and any item with neither content nor an excerpt. WordPress’s own “Discourage search engines” setting is deliberately ignored, since staging sites routinely have it on. Sites can exclude further content by category in Settings → Ask My Brand, or by any WP_Query criterion with the epubamb_feed_query_args filter.

Only the feed’s own URL serves content. WordPress also exposes /<slug>/feed/ask-my-brand/, /comments/feed/ask-my-brand/ and archive forms such as /category/news/feed/ask-my-brand/ as a side effect of registering a named feed, and the same applies to WordPress query variables added to the feed URL — &cat=, &year=, &tag=, &s= and the like. All of these return 404. Use the documented category and content_type parameters instead. The one exception is the posts page of a site with a static front page (for example /blog/feed/ask-my-brand/), which serves the same feed as the canonical URL.

For developers

Posts and pages are in the feed out of the box. Other content types — WooCommerce products, or any custom post type — are opt-in, because this feed transmits content to a third party and installing a plugin must never widen that payload on its own.

Add one with the epubamb_content_types filter, which maps a post type to the label the feed reports:

add_filter( 'epubamb_content_types', function ( $types ) {
    $types['product'] = 'Product';
    return $types;
} );

Products then appear in the feed as <askmybrand:content_type>Product</askmybrand:content_type>, and callers can request ?content_type=Product to fetch only those. Several types may share a label — label a news type Article and ?content_type=Article returns posts and news together. Labels may not contain a comma.

Notes:

  • The label is part of the feed’s public contract — the indexer sees it, and callers filter on it. Use a stable literal, not a translated post type label, or the value changes with the site’s locale.
  • Register the post type by the init action, as WooCommerce and most plugins do. A type registered later is dropped from the map, because the feed refuses to advertise a type it cannot query.
  • Types you add inherit the feed’s behaviour: full content, password-protected items excluded, and the content_type element on every item.
  • Filtering by category returns posts only. Pages and most custom post types do not carry the category taxonomy, so they are excluded from a category-filtered request and are not listed in <askmybrand:contentTypes> for it.
  • A post type that is not registered — its plugin deactivated, say — is dropped from the map, so the feed never advertises a type it cannot return.
  • Settings → Ask My Brand lists the content types currently in the feed, which is the quickest way to confirm a filter took effect.

Changing what the feed selects: epubamb_feed_query_args

Settings → Ask My Brand covers excluding whole categories. For anything else — a different taxonomy, a meta key, an exclusion built at runtime — filter the WP_Query arguments the feed runs on:

add_filter( 'epubamb_feed_query_args', function ( $query_args, $request ) {
    unset( $query_args['category_name'] ); // drop the slug allow-list
    $query_args['cat'] = '-12,-15';        // exclude categories 12 and 15
    return $query_args;
}, 10, 2 );

$query_args is the array handed to `WP_Query`; `$request` is the parsed feed request (`paged`, `per_page`, `categories`, `post_types`), so a filter can branch on what was asked for.

Notes:

  • totalPosts and totalPages are read from the feed query itself, so they always describe what the feed actually returns.
  • Use negative cat IDs to exclude, not category__not_in. They look equivalent and are not: WordPress resolves negative cat IDs with include_children => true, so excluding a parent category also excludes its children, while category__not_in leaves the children in the feed.
  • An exclusion leaves content carrying no categories alone — pages and most custom types stay in the feed. An inclusive filter (category_name, category__in, positive cat IDs) drops them, and <askmybrand:contentTypes> reflects that.
  • post_status, has_password and the noindex / “Hide from Ask My Brand” exclusions are re-applied after the filter and cannot be overridden. This feed transmits content to a third party, so no filter gets to widen it past published, unprotected content the site is willing to have indexed. A meta_query you set is kept and combined with these exclusions.
  • paged and per_page are not filterable: they are the URL’s pagination contract, not a statement about which content the feed carries.
  • The filter runs once per request. The channel metadata (contentTypes, categories) is read from the query that actually ran, so it always describes the items.
  • Categories excluded in Settings → Ask My Brand (and their children) are enforced after the filter and cannot be overridden — not by your own cat, nor by a tax_query with an OR relation.
  • A tax_query you set is kept as one nested group, so its relation applies among its own clauses only. The category parameter (and any category_name or cat you set) is ANDed with it, not added as one more OR alternative.

Installation

  1. Upload the ask-my-brand-feed-by-epublishing folder to /wp-content/plugins/
  2. Activate the plugin through the ‘Plugins’ menu in WordPress
  3. Go to Settings → Ask My Brand
  4. Click “Generate Token” to create your secure feed URL
  5. Share the feed URL with Ask My Brand for content ingestion

Changelog

1.6.0

  • Pages are now included alongside posts, each item labelled with ; turn pages off in Settings → Ask My Brand
  • New settings: exclude categories, and a ‘Hide from Ask My Brand’ checkbox on the edit screen
  • Content marked noindex in Yoast SEO or Rank Math, password-protected content and WooCommerce Cart/Checkout/My Account pages are no longer sent
  • Items are ordered by last change, so incremental syncs pick up edited and scheduled content
  • Your existing feed URL and token keep working after the update
  • Featured image thumbnails and captions are still included
  • Requires WordPress 5.5 or later

1.5.2

  • Upgrades no longer switch pages off. 1.5.1 turned them off for any upgraded site whose settings had never been saved, which also caught 1.4 sites that were already sending pages. A site that has not saved its settings now keeps pages on; tick or untick “Include pages in the feed” to choose
  • The plugin now records its installed version, so later upgrades can tell which release a site is coming from
  • A request without a valid token is rejected with 401 before WordPress runs its query, including HEAD requests and feed URLs that would otherwise 404
  • Settings → Ask My Brand lists the published items the feed is leaving out, with a link to edit each and the reason (noindex, excluded category, pages turned off, empty content, hidden)
  • “Regenerate Token” now asks for confirmation first, since a new token stops the current feed URL working

1.5.1

  • Items with neither a body nor an excerpt (a blank page or post) are no longer sent, and are not counted in totalPosts
  • <style> and <script> blocks are stripped from item content
  • Fixed: a developer filter whose tax_query used an OR relation let posts in categories excluded in Settings back into the feed, and turned the category parameter into one more alternative — the feed served everything while still reporting the requested categories
  • Sites upgrading from 1.3 or earlier start with pages switched off; new installs start with them on
  • Fixed: a developer filter that set its own cat silently dropped the categories excluded in Settings
  • Fixed: saving the settings discarded exclusions on categories the page did not list, such as another language’s categories under Polylang or WPML
  • Fixed: the feed answered conditional requests with a site-wide ETag before checking the token, so a tokenless request could get a 304 and a crawler could miss changes. It no longer sends or honours ETag / Last-Modified
  • Fixed: an HTML entity in a category or label (such as &eacute;) produced invalid XML
  • Fixed: the Feed URL on the settings page was wrong on sites using Plain permalinks
  • contentTypes now accounts for any inclusive taxonomy filter (tags, product categories, EXISTS), not only categories
  • Requires WordPress 5.5 or later

1.5.0

  • Content marked noindex in Yoast SEO or Rank Math is no longer sent, including content that is noindex only through its post type’s default. An item explicitly set back to index is still sent. Rank Math is honoured only once it is registered or its setup wizard skipped, since until then it outputs no robots tag
  • New “Hide from Ask My Brand” checkbox on the edit screen keeps a single post or page out of the feed
  • WordPress’s “Discourage search engines” setting does not affect the feed
  • WooCommerce’s Cart, Checkout and My Account pages are never sent
  • Items are now ordered by when they last changed — the later of last edit and publication — rather than publication date, so incremental syncs pick up edited content and scheduled posts. <lastModified> reports the same value
  • Fixed: on a site with a static front page, the posts-page feed (/blog/feed/ask-my-brand/) prepended sticky posts, overfilling the page and exposing the titles of password-protected sticky posts
  • Feed URLs other than the feed’s own — single-item, comment and archive forms, or the feed URL with raw WordPress query variables — now return 404. A single-item URL previously served that item in full even when its content type was switched off
  • totalPosts and totalPages now come from the feed query itself instead of a second count query
  • Content type labels containing a comma are ignored. Several types may share a label, and content_type then returns all of them

1.4.0

  • Fixed: the pagination block was emitted on archive feed URLs such as /category/news/feed/ask-my-brand/, and whenever a raw WordPress query variable (&cat=, &year=, &s=) was added to the feed URL. The items were narrowed but the totals described the whole site, so an indexer following totalPages walked into 404s. The block is now emitted only where the totals are known to match
  • Settings → Ask My Brand now carries the Ask My Brand logo and brand colours
  • New Exclude Categories setting: tick categories in Settings → Ask My Brand to keep them out of the feed. Excluding a parent category also excludes its children, and the pagination totals account for it
  • New epubamb_feed_query_args filter: developers can change any WP_Query argument the feed selects on — exclusion lists, other taxonomies, meta queries. The same arguments drive the pagination count, so the totals can never describe a different set than the feed returns
  • post_status and has_password are enforced after that filter, so no filter can widen the feed past published, unprotected content — this also stops a logged-in administrator hitting the feed URL from seeing private posts
  • Pages are now included in the feed alongside posts, and can be turned off in Settings → Ask My Brand
  • New <askmybrand:content_type> element on every item — Article for posts, Page for pages
  • New content_type parameter for filtering by type
  • New epubamb_content_types filter: developers can add custom post types to the feed, so a WooCommerce product can be indexed as <askmybrand:content_type>Product</askmybrand:content_type> and fetched with ?content_type=Product
  • Custom types are opt-in by design — installing a plugin never widens what the feed transmits to Ask My Brand on its own
  • Types added this way inherit the feed’s behaviour: full content, password-protected items excluded, and a content_type element on every item
  • Settings → Ask My Brand now lists the content types currently in the feed, which confirms at a glance whether a filter took effect
  • Documented the feed’s XML format in readme.txt: the askmybrand namespace, the pagination block and its children, the per-item elements, how to crawl, and what is excluded
  • Fixed: per_page and category were accepted but silently ignored, so the feed always returned 10 unfiltered items while the pagination metadata reported the requested values
  • Fixed: password-protected posts and pages were sent in full; they are now excluded
  • Fixed: content containing ]]> produced malformed XML

1.3.3

  • Emit media:thumbnail and media:description per item (Yahoo Media RSS namespace) so consumers can ingest featured-image URLs and captions alongside content.

1.3.2

  • Add Docs, Improve Performance

1.3.1

  • Add Icons

1.3.0

  • Initial release