WordPress posts tagged by AI, with missing tags created

WordPress posts tagged by AI, with missing tags created

An n8n workflow that asks an AI model for three to five tags for an article. It creates in WordPress the tags the site lacks and returns the ids of all of them. A demo flow runs it on the items of an RSS feed and creates a WordPress post with the tags.

What is inside

• workflow.json: the workflow as n8n exports it, thirty-one nodes: twenty-two that do the work and nine sticky notes. It holds no credential.
• README.md: these notes.

What we changed from the original copy

• The credential links are gone. Select your own for OpenAI and WordPress.
• The two OpenAI Chat Model nodes named no model, so n8n's default gpt-5-mini would run. OpenAI's deprecations page, read on 12 November 2026, lists it as shut down on 11 December 2026, with gpt-5.6-terra as the replacement. Both now name gpt-5.6-terra. The chain and the auto-fixing parser call no tools.
• Auto-Tag Posts in WordPress called a workflow by an id from the original instance, and that id was the file's own. It now reads ={{ $workflow.id }}, so the file calls itself after you import it.
• RSS Feed Trigger had no feed address. It now holds <YOUR_RSS_FEED_URL>.
• Three HTTP Request nodes pointed at www.example.com. They now hold <YOUR_WORDPRESS_SITE>.
• The two nodes that list tags asked for no page size. WordPress's reference for tags, read on 12 November 2026, gives per_page a default of 10. On a site with more tags, the check for missing ones would miss existing tags, and creating them again would fail. Both nodes now ask for per_page=100, the most the handbook allows.
• The demo post's text said "postvia". It now says "post via".
• The sticky notes were reworded, and one that held only a profile link was cut.
• Nothing else.

Before you start

• An n8n instance, on n8n Cloud or your own server.
• An OpenAI API key.
• A WordPress site with a credential in n8n: a username, its password and the site address. A WordPress application password works for this.
• For the demo, an RSS feed whose articles you may reuse.

Set it up

1. In n8n, create a workflow, open its menu and choose Import from File. Pick workflow.json.
2. Add an OpenAI credential and select it in OpenAI Chat Model and OpenAI Chat Model1.
3. Add a WordPress credential and select it in GET WP tags, POST WP tags, GET updated WP tags and Wordpress. The three HTTP Request nodes use it as their predefined credential.
4. In those three HTTP Request nodes, replace <YOUR_WORDPRESS_SITE> with your site's domain.
5. In RSS Feed Trigger, replace <YOUR_RSS_FEED_URL> with the address of a feed.
6. Save the workflow and leave Auto-Tag Posts in WordPress as it is. It finds this workflow on its own.
7. Try it on a test site first. The demo creates a post on the site.
8. Activate the workflow.

How a run goes

The demo:

1. RSS Feed Trigger reads the feed every minute and starts a run for each new item.
2. Generate tags for article asks for three to five tags in title case for the item's content. Structured Output Parser expects a JSON object with a tags array, and Auto-fixing Output Parser asks the second model to repair a reply that does not fit.
3. MOCK article collects the title, the content, the categories and the tags.
4. Auto-Tag Posts in WordPress runs this workflow once for each item, starting at Demo Usage in Another Workflow.
5. Return article details joins the returned tag_ids with the title and content, and Wordpress creates a post called "Demo tagging post" followed by the title, with those tags.

The tagging steps:

1. Loop over articles hands in one article at a time, and SET initial record keeps it.
2. GET WP tags lists the site's tags, and Combine slugs gathers their slugs.
3. Return missing tags turns each tag from the AI into a slug, lowercase with hyphens for spaces, and lists the ones the site lacks.
4. If checks that list. When nothing is missing, the flow goes to GET updated WP tags. Otherwise Split Out and POST WP tags create each missing tag with its slug and a title-case name, and then GET updated WP tags runs.
5. Keep matches keeps the tags whose slug is among the article's, and Combine tag_ids gathers their ids.
6. When the loop ends, the ids go back as tag_ids.

Change the look

• The number and style of tags: the prompt in Generate tags for article.
• The case of slugs and names: Return missing tags, Keep matches and the name expression in POST WP tags.
• The model: both OpenAI Chat Model nodes.
• Use it in your own flow: remove Demo Usage in Another Workflow and put the tagging steps inline, or call the file as a sub-workflow.

Good to know

• The tag lists stop at 100 tags. A site with more needs paging added to the two GET nodes.
• The flow makes a slug by lowercasing and swapping spaces for hyphens. WordPress cleans slugs further, so a tag with punctuation may not match, and its id is left out.
• The demo copies the items of the feed into posts on your site. Use a feed you may republish.
• The file sets no status on the demo post, and WordPress's reference gives no default for creating one. Check the status of the new post.
• Each article costs up to two model calls.
• This copy was not run against WordPress or OpenAI.
• The article text goes to OpenAI, and the tags and the demo post go to WordPress.
• The JSON holds no credential.