The metadata you just graded arrives with the post.
Most integrations push a title and a body and quietly drop everything else. That is how a piece scores well in the tool and appears in search with a snippet the theme truncated out of its first paragraph.
The last hundred metres is where the work leaks out.
A team spends real effort on a meta description, an FAQ block and structured data, because those are the things that decide how the piece looks in a result and whether an assistant can quote it. Then the publish step sends a title and some HTML, and all of it stays behind in the tool that graded it.
Nobody notices for months, because the post is live and looks fine. What is missing is only visible in a search result, and by then there are forty of them.
A publish that loses the metadata makes every check that graded it decorative.
Connect once, then approve or schedule, and everything travels.
- 1
Connect with an application password
Per-application, revocable on its own, never your account password. Docket checks the account can publish before accepting it, and stores the credential encrypted.
- 2
The publish claims the article
An atomic claim in the database means two clicks, or a click racing the scheduler, cannot produce two posts on your live site.
- 3
Body, image, schema, metadata
The image is uploaded to your media library. The structured data carries the real permalink. The meta fields follow in a second request that cannot take the publish down with it.
- 4
Now, scheduled, or only on approval
Per website, and the default requires a human. Governance rules can also block a publish outright on brand grounds, with the reason shown in the editor.
A publish you do not have to check afterwards.
Every line here describes something the software actually does. If one of them turns out not to be true of your setup, that is a bug and we want to hear about it.
Two concurrent publishes cannot create two posts.
The claim is a conditional update PostgreSQL evaluates atomically, so exactly one caller proceeds. Timestamped, so a crash mid-publish clears itself within minutes rather than locking the article.
Yoast and Rank Math are both supported without asking which you use.
Both sets of field names are sent. A site with neither rejects that one request and keeps the post, and the log records what did not land.
Pushing a live article back to draft does not erase its publish date.
That date anchors decay scoring and the outcomes loop. It used to be wiped by a draft push, and there was nothing to recover it from.
Structured data uses the real URL.
On a first publish WordPress decides the permalink, so the schema is written in a second pass once it is known rather than guessed from the slug, which is wrong on any non-default permalink structure.
Every publish is in the activity log.
What went out, when, to which post ID, at which URL, whether the image and the metadata landed. Per website, and readable by everyone on the team.
What it does not do
- WordPress only today. Webflow, Shopify, Ghost and a generic webhook are scaffolded in the data model and not finished, and we would rather say that than imply a roadmap date.
- It does not create or assign categories, for the reason above.
- It does not manage redirects, internal link maintenance across your existing site, or anything about pages Docket did not write.
- It cannot publish to a WordPress that is not reachable from the internet, or one where the REST API is disabled.
Answers, Not Hedging.
The parts either side of this one.
Twenty minutes, your website on the screen.
Not a slide deck. We run the checks against your actual domain before the call and spend the time on what came back. If it turns out you do not need this, we will say so.