Retresco main entity toggle

content/guides/integrations/retresco.md

@@ -36,12 +36,6 @@ For the test environment it is recommended to either disable the Retresco integr
 
 Each project also needs to be configured to use the Retresco integration. You must set `enabled: true` to activate the feature, and provide the Retresco API endpoint and credentials.
 
-You can also optionally provide the text extraction "matches". `titleMatches`, `supertitleMatches`, and `teaserMatches` are arrays of `<component-name>.<directive-name>` strings which can be used to extract the text content and deliver them to the Retresco API in the relevant fields within the request payload. Only the first match found for each field will be used, but multiple values can be provided so that many content types can be supported.
-
-The default behaviour is to analyse the text on the idle event, typically 5 seconds after a user stops typing, so that the tags are always up-to-date with the content. If you would like to prevent text analysis while a user is working on a document it is possible to set `enableLiveAnalysis: false`. Once disabled the only way the tags will be updated is by the user clicking the refresh button in the metadata form, or at the time of publication.
-
-The Retresco API can have issues if the text sent for analysis is too long (more than 10.000 characters). To avoid errors with longform articles, you can optionally set a `maxTextLength`, this will limit the number of characters of text being sent in the request.
-
 ```js
 {
   // ...
@@ -59,6 +53,7 @@ The Retresco API can have issues if the text sent for analysis is too long (more
           }
         },
         enableLiveAnalysis: true,
+        enableMainEntityOverride: true, // Optional
         titleMatches: ['header.title'],
         supertitleMatches: ['header.catchline'],
         teaserMatches: ['p.text'],
@@ -73,6 +68,26 @@ The Retresco API can have issues if the text sent for analysis is too long (more
 }
 ```
 
+#### Text Extraction
+
+You can also optionally provide the text extraction "matches". `titleMatches`, `supertitleMatches`, and `teaserMatches` are arrays of `<component-name>.<directive-name>` strings which can be used to extract the text content and deliver them to the Retresco API in the relevant fields within the request payload. Only the first match found for each field will be used, but multiple values can be provided so that many content types can be supported.
+
+#### `enableLiveAnalysis`
+
+The default behaviour is to analyse the text on the idle event, typically 5 seconds after a user stops typing, so that the tags are always up-to-date with the content. If you would like to prevent text analysis while a user is working on a document it is possible to set `enableLiveAnalysis: false`. Once disabled the only way the tags will be updated is by the user clicking the refresh button in the metadata form, or at the time of publication.
+
+#### `maxTextLength`
+
+The Retresco API can have issues if the text sent for analysis is too long (more than 10.000 characters). To avoid errors with longform articles, you can optionally set a `maxTextLength`, this will limit the number of characters of text being sent in the request.
+
+#### `enableMainEntityOverride`
+
+Retresco marks the strongest tags in a document as "main entities". By default this state is stored but not editable by users in the editor. Set `enableMainEntityOverride: true` to show a star toggle on each tag so editors can promote or downgrade main entities themselves ({{< added-in "release-2026-07" >}}).
+
+{{< info >}}
+Before enabling this option, check with Retresco that main entities are enabled on your setup.
+{{< /info >}}
+
 ### Content Type
 
 You will need to add the metadata plugin for all content types that you would like to analyze. You may provide any handle for the Retresco metadata field. The `li-retresco` type is essential.
@@ -108,12 +123,23 @@ The Retresco entities will be stored in the document's metadata using the metada
       score: 16.924834941594718, // Only when not user-added
       userAdded: false,
       inappropriate: false, // Entity removed (but still visible in the UI with strikethrough)
-      isMain: true // Marks main entities ({{< added-in "release-2024-09" >}})
+      isMain: true, // Marks main entities ({{< added-in "release-2024-09" >}})
+      userOverrideIsMain: true // Set when an editor changed the main state ({{< added-in "release-2026-07" >}})
     }
   ]
 }
 ```
 
+### Main Entities
+
+{{< added-in "release-2026-07" block >}}
+
+Retresco marks the strongest tags in a document as main entities in the `isMain` field. When `enableMainEntityOverride` is enabled on the project, editors can override this verdict with a star toggle on each tag.
+
+{{< img src="./retresco-main-entities.png" alt="Star toggle to mark a Retresco tag as a main entity" width="600" caption="Editors mark a tag as a main entity with the star toggle." >}}
+
+Toggling a tag stores the new value in `isMain` and sets `userOverrideIsMain: true`. Once set, the tag stays even if Retresco no longer detects it, and re-enrichment no longer overwrites the editor's choice.
+
 ## Re-enrich Documents
 
 {{< added-in "release-2023-03" block >}}

content/operations/releases/release-2026-07.md

@@ -280,6 +280,22 @@ Which options are available depends on the dashboard's asset type and your confi
 
 For more information, see the [Card Tags and Display Settings]({{< ref "/guides/media-library/media-library-setup/#card-tags-and-display-settings" >}}) documentation.
 
+### Retresco Main Entity Toggle
+
+Retresco marks the strongest tags in a document as "main entities". Livingdocs stored this state but never showed it. Editors can now see and change it themselves. A star toggle on each tag promotes or demotes it as a main entity.
+
+{{< img src="release-2026-07-retresco-main-entities.png" alt="Star toggle to mark a Retresco tag as a main entity" width="600" caption="Editors mark a tag as a main entity with the star toggle." >}}
+
+The feature is opt-in. Enable it per project with the new `enableMainEntityOverride` setting in the Retresco project config.
+
+{{< info >}}
+Before enabling this option, check with Retresco that main entities are enabled on your setup.
+{{< /info >}}
+
+The Retresco and iMatrics tag components also got a design refresh. Tags are now sorted by type and then alphabetically, and a new indicator marks tags an editor added or changed.
+
+For more information and configuration details, see the [Retresco integration]({{< ref "/guides/integrations/retresco" >}}) documentation.
+
 ### Keep Inbox Items After Dropping Them Into a Document
 
 By default, an inbox item is removed from the inbox as soon as it is dragged into a document. Some workflows call for the opposite. An editor researching images for an article might want to keep them in the inbox while trying out different options during production. For those setups the item should remain in the inbox after it has been placed.

content/reference/document/metadata/plugins/li-retresco.md

@@ -31,7 +31,8 @@ storageFormat: |
         score: <Number>,
         inappropriate: <Boolean>,
         userAdded: <Boolean>,
-        isMain: <Boolean> // {{< added-in "release-2024-09" >}}
+        isMain: <Boolean>, // {{< added-in "release-2024-09" >}}
+        userOverrideIsMain: <Boolean> // {{< added-in "release-2026-07" >}}
       }
     ]
   }