Skip to content

Automation · Content and publishing

Publish a WordPress article with its image, its metadata and its schema

How an AI agent publishes to WordPress: it detects the editor and the SEO plugin, converts to blocks, uploads the featured image and verifies the result.

How do I publish to WordPress without half an hour of pasting blocks and filling metadata?

The Met picks the topic on real demand data, detects which editor and which SEO plugin that WordPress runs (because they are not all the same, and publishing as if they were produces a malformed article), converts the text into editor blocks, uploads the image and sets it as featured, publishes, writes the metadata and the structured data, and then downloads the public HTML to verify that everything landed where it should. That last step is what separates "I published it" from "it is published correctly".

What starts it

When someone asks for it

It runs when the team asks for it, in chat or with the button on the task. Publishing content unattended is the fastest way to fill a site with articles nobody reviewed.

A manual task: someone on the team runs it with a button, or asks the Met for it in chat.

Who runs it

Met for content and SEO

Social team

It has no seeded name in the product: it is an internal marketing Met. It needs the site tools and the demand tools at the same time, which is the combination that gets the article written about something people actually search for.

What you need connected

6 steps · 16 tools

The procedure, step by step

Each step shows the tool that runs. They belong to the real connector: if one of them stopped existing, this page would not build.

  1. 1

    Pick the topic on real demand, not on instinct

    Related queries return what people are searching around a term, and interest over time says whether it is a trend on the way up or one that already passed. Web search shows what is published today on the subject. The three together answer the only question that matters before writing: is this article worth it?

  2. 2

    Find out which WordPress you are working against

    Editor detection says whether the site runs blocks or the classic editor, and that completely changes how the content has to be formatted. SEO plugin detection reveals whether there is Yoast, Rank Math, SEOPress or none at all. Assuming is what produces articles with the HTML pasted as plain text, and nobody notices until somebody opens the site.

  3. 3

    Convert to blocks and prepare the image

    Conversion turns the Markdown text into WordPress editor blocks, with well formed headings and paragraphs. The image is uploaded to the media library from a URL and set as featured, which is the one that shows in the blog listing and on social networks when somebody shares the link.

  4. 4

    Publish, or leave a draft if somebody reviews

    Creation publishes straight away by default; leaving it as a draft is an explicit decision for when there is editorial review. The response carries the public link, which gets quoted verbatim instead of reconstructed. A draft is published later by changing only its status.

  5. 5

    Write the metadata and the structured data

    On sites with the corresponding plugin, the SEO metadata (title, description, canonical, Open Graph) is set from here, and the JSON-LD is injected into a field the schema plugin reads and publishes in the head. Both steps depend on that plugin being installed: without it the article still publishes, just without this layer.

  6. 6

    Verify the public HTML, do not trust the response

    Validation downloads the published page, checks that the JSON-LD is present and parseable, and returns a direct link to Google's rich results test. It is the step nobody does by hand and the one that catches that the plugin was not active, that the cache served an old version or that the schema came out malformed.

Before you start

What you need ready

The Met solves none of these for you. If one is missing, the automation stops there.

The limits

What this automation does NOT solve

Better said here than discovered halfway through the rollout.

FAQ

Questions about this automation

Because the automation runs against different sites and the same Met handles several of them. Detection costs one call, can be remembered for a day, and prevents the quietest failure of all: an article published with the HTML visible on the page.

Technically yes, and the trigger for this automation is manual precisely because it is a bad idea. A site that publishes on its own piles up articles nobody read before they went out and nobody reads after, and you pay for that in credibility.

The article stays published, since that already happened, and validation reports what is missing. Usually it is the site cache or a deactivated plugin, and both are fixed without rewriting anything.

Keep going

Other automations built on the same tools

Want this automation running on your own data?

We will show it to you with your accounts connected, not with a canned demo.