Hydrogen SEO 0.0.27: Site Verification, IndexNow and a 404 Monitor

Hydrogen SEO 0.0.27 adds three screens to the free plugin: Webmaster Tools, for the verification codes Google Search Console, Bing and four other services ask for; Instant Indexing, which sends changed URLs to the search engines that use IndexNow; and a 404 Monitor, which records the broken URLs real requests hit, without storing IP addresses. None of them needs an account. The release also fixes how the robots.txt and llms.txt editors save text, how schema stores postal codes, how quickly the sitemap reflects settings, and several things in the editor's SEO sheet.

Update from Plugins → Installed Plugins as usual. Two of the fixes change stored data on update, and one is worth a minute of checking afterwards; the upgrade notes are at the end.

Webmaster Tools

Every search console wants proof that you control the site before it shows you anything. The quickest proof is a <meta> tag on your home page, and until now Hydrogen SEO gave you nowhere to put one except a code snippet.

Settings → Webmaster Tools has a field for each of six services:

• Google Search Console (google-site-verification)
• Bing Webmaster Tools (msvalidate.01)
• Yandex Webmaster (yandex-verification)
• Pinterest (p:domain_verify)
• Baidu Search Resource Platform (baidu-site-verification)
• Naver Search Advisor (naver-site-verification)

Paste either the code or the whole tag the service shows you. A valid code is saved as soon as you paste it; from a whole tag, only the code inside it is kept, and the field shows the exact tag it will print underneath. Anything that is not a plain token is not saved, with a message saying so; a code saved before stays in place, and the other services save as normal, so a half-copied tag or a DNS record pasted into the wrong box never reaches your pages. The tags are printed in the head of every front-end page, inside the Hydrogen SEO block, so you can press Verify in the service straight after pasting.

Switching from another SEO plugin? Import SEO Data now brings its verification codes over into this screen.

Keep the codes there after you verify. Google and Yandex both recheck ownership from time to time, and a missing tag eventually costs you access. Setup notes for each service, and the cases where a DNS record is the better method, are in the Webmaster Tools guide.

Instant Indexing (IndexNow)

IndexNow is an open protocol for telling search engines a URL has changed. A site hosts a key in a text file to prove it owns the domain, then posts lists of changed URLs; the participating engines share every submission among themselves. According to indexnow.org it is supported by Bing, Naver, Seznam.cz, Yandex and Yep. Google does not take part.

Settings → Instant Indexing is off until you switch it on, because it sends your URLs to an outside service. Once on:

• Hydrogen SEO generates a key and serves the key file at https://yoursite.com/<key>.txt, with Plain permalinks too. Some servers answer .txt addresses themselves on Plain permalinks, so the screen warns you and points to Check key file.
• When a post of a type you selected is published, updated while published, or unpublished, trashed or deleted, its URL is queued; after a slug change, the old address is sent too. Only posts that are indexable, or were until this change, are announced: a post that is noindexed, password-protected or canonicalised to another page sends nothing when it is published, edited, trashed or deleted, and a post noindexed by the change you just made is sent once so engines drop it.
• About 30 seconds after the first change, everything queued goes out in the background as one request to the shared endpoint at api.indexnow.org, so a burst of edits is one submission.
• Sending happens only on production sites. On a site whose WordPress environment type is local, development or staging, the screen says so and nothing goes out.
• A send that fails because IndexNow cannot be reached, does not answer within 3 seconds, or answers 429 or a server error is retried automatically about 10 minutes later, up to three attempts per URL. A 400, 403 or 422 answer is final, since it needs a fix on your side.
• Recent submissions lists the last 50 requests with their time, URL count and the IndexNow response in plain words: 200 received, 202 received with the key still being validated, 403 key not valid, 422 URL not on this site, 429 too many requests.
• A Submit URLs box takes URLs on your site one per line, Check key file confirms the key file is reachable, and New key replaces the key.

It will not make anything appear in Google, and a 200 means received, not indexed. For a site that changes daily it shortens the wait for Bing and the others; for a site that publishes monthly it is harmless and mostly unnoticeable. The Instant Indexing guide covers each response and how to confirm delivery in Bing Webmaster Tools, and IndexNow on WordPress goes into what to expect from it.

404 Monitor

Search Console shows the 404s Googlebot finds. It does not show the reader who followed a mistyped link from a forum, or the newsletter from last spring pointing at a post you have since renamed. Tools → 404 Monitor does.

Switch it on and every front-end page request that still ends in a 404 after your redirects have run is recorded, grouped by URL: the path without its query string (except on the home page, where ?p=123 is what identifies the page), the hit count, when it was first and last seen, and the last referrer and user agent. IP addresses and query strings are not stored, which also means a scanner varying the query string cannot add a row per request. Requests to wp-admin, the REST API, AJAX and WP-Cron, and POST requests, are ignored. Entries live in their own database table, URLs not hit within the retention period you choose (7, 30 or 90 days, 30 by default) are removed, and the log is capped at 5,000 addresses.

On the screen you can search, sort by hits or last seen, delete rows or clear the log, and turn any row into a 301 or 302 redirect that lands in the existing redirection manager. URLs that already have a redirect are marked. The monitor is described in its guide; how to read what it collects is in Your 404 Log Is a To-Do List.

Fixes

robots.txt editor. Files are saved exactly as typed: one directive per line, with & and percent-encoded characters such as %C3%A9 kept intact. Earlier versions saved the whole file on one line, which crawlers read as a single broken rule. Updating repairs a file stored that way once, putting each directive back on its own line. The repair never turns text after a # into rules, so a commented-out rule stays a comment, and it changes & written by earlier versions back to &.

llms.txt editor. Links containing percent-encoded characters keep them, so URLs with accented or non-Latin slugs stay valid. In both editors, a < that does not start a tag (as in <50ms) and markdown autolinks such as <https://example.com/> survive saving.

Redirections. A redirect whose source address contains non-Latin or accented characters, such as /café/, now matches requests for it, whether the rule was typed with the characters themselves or percent-encoded.

Schema screen. Number-like text stays text. A postal code such as 04101 keeps its leading zero, and coordinates such as 40.7128 keep their decimals. Address parts in the Organization schema are printed as text.

Sitemaps. The sitemap reflects site-wide settings on the next request: noindex for a post type or taxonomy, noindex for empty archives, Strip Category Base, the permalink structure, category base and tag base, and "Discourage search engines". It used to wait for its cache to expire.

Per-post schema in the SEO sheet.

• FAQ, HowTo, Local Business and other per-post schemas now print the nested types they need, such as Question and Answer inside an FAQ and PostalAddress inside an address, and use each field's default when you leave it blank.
• Schemas saved on a post are listed again after a reload, and saving the sheet no longer removes them.
• Repeater and group fields keep everything you type.
• A subtype you choose, such as BlogPosting instead of Article, is the type printed.
• Defaults only fill in groups and rows you entered something in, so no rating without a rating value or offer without a price is printed.
• Smart tags such as %date% and %category% in per-post schema fields are saved intact.

Schema author fields. For a post without an author, author fields are left empty instead of pointing at a bare /author/ address.

Editor. The Media Library opens and works from the SEO sheet, and the image you choose there is saved. The editor's menu entry and toolbar button read "Hydrogen SEO", and so does the HTML comment around the plugin's head output.

Upgrade notes

1. Read your live robots.txt once. If an earlier version saved it on one line, the update has split it back into lines. Open Tools → Robots.txt Editor → View Robots.txt and check the rules are the ones you meant.
2. Re-save number-like schema values. A postal code or coordinate saved by an earlier version may already have lost its leading zero or decimals. Check the address and coordinates on the Schema screen and save again.
3. Check per-post FAQ and HowTo schema. If you added them before updating, open each post's SEO sheet, confirm the questions or steps are complete, save, and test the page in Google's Rich Results Test or the Schema Markup Validator. FAQ rich results are gone from Google for every site since 7 May 2026, so the markup is for machines that read it rather than for a visual result.

The three new screens are opt-in or empty until you use them, so nothing changes on your pages until you fill in a code or switch a feature on.