Why don’t robots.txt and llms.txt show your panel changes?

SEO/GEO Files Generator is a WordPress plugin that generates and serves robots.txt, llms.txt and llms-full.txt from a single panel. It is made by BeeClear, a developer of WordPress plugins and smartphone apps. Sometimes you save new rules, yet /robots.txt keeps showing the old content. The usual culprit is a physical file on the server that quietly takes over from the panel settings. Below you will find the symptoms, their causes and the ways the plugin helps you spot them.

Why does robots.txt show old content after you save the settings?

Robots.txt shows old content after you save the settings mostly when a physical file with that name sits in the WordPress root. The web server returns that file instead of the response WordPress would generate. Your new rules land in the database, but no crawler ever sees them. SEO/GEO Files Generator builds robots.txt dynamically once you tick “Serve generated robots.txt through WordPress”. A file on disk still wins over anything WordPress produces.

A server administrator may have uploaded that file by hand. Just as likely, the generator itself wrote it earlier with the “Save & write robots.txt file” button. Either way the symptom looks the same: the preview in the panel shows the new rules, while the live address shows the old ones.

How does the plugin detect a physical robots.txt in the WordPress root?

The plugin detects a physical robots.txt in the WordPress root every time you open the robots.txt screen. When the file exists, a warning below the form says it is served instead of your settings. The message also tells you whether the file content differs from the preview. The comparison ignores line ending differences and trailing spaces, so formatting alone raises no alarm.

Different content counts as an error with a “Blocked” badge. Matching content only triggers a warning with a “Check” badge. Crawlers then read the right rules, but every later change means writing the file again. Next to the preview you also get an “Open live robots.txt” link to the version served at your site address.

When can you delete a physical robots.txt from the panel?

You can delete a physical robots.txt from the panel only when the plugin wrote it. Such a file starts with the header “# robots.txt generated by BeeClear SEO/GEO Files Generator”. For that file a “Delete physical robots.txt” button appears, with a confirmation prompt. A file without that header counts as hand-made, and the generator leaves it alone. The message then suggests deleting it directly on your hosting. Here is how the plugin treats the file depending on where it came from:

Situation on the server	What the panel shows	What happens on a plugin update
file written by the plugin, content matches the settings	warning and a delete button	the file stays in place
file written by the plugin, content is outdated	error and a delete button	the file is removed and WordPress serves the current settings
hand-made file	warning or error, no button	the file stays in place
no file	no warning	WordPress keeps serving the saved settings

Automatic cleanup works from version 2.5.1 and handles the most common case. An outdated copy written by the generator no longer hides your new rules. A hand-made file stays your call, since the plugin cannot tell its origin.

Why does llms.txt link to a different domain than yours?

An llms.txt linking to a different domain is a typical sign of a file written to disk once. A physical file keeps the content from the moment of generation and ignores the address you open it under. Since version 2.4.0, SEO/GEO Files Generator serves llms.txt and llms-full.txt dynamically by default. The links then point to the domain the site is opened on: production, a staging copy, a tunnel or a local port. You choose the mode in Global settings, under “llms delivery”.

Dynamic mode caches the output per domain. Every regeneration refreshes that cache. Writing to disk is still available as the “Physical files written to the WordPress root” option. Choose it deliberately: a file on disk carries one set of links whatever the domain.

What should you do when an old llms.txt hides the dynamic version?

An old llms.txt hides the dynamic version the same way a physical robots.txt hides your settings. The server returns what is on disk, and WordPress never gets a chance to answer. The plugin flags this in the status line on the llms.txt and llms-full.txt screens. The message states plainly that a file on disk is served instead of the dynamic output.

On every regeneration in dynamic mode the generator tries to delete that file. It recognises its own files by the UTF-8 BOM marker at the very start. A file without that marker counts as hand-made and stays where it is. You will then see a note that llms.txt was refreshed, with a warning that the server still returns the file from disk. Delete it from the server, and the dynamic version starts working.

Why doesn’t a page edit reach llms.txt straight away?

A page edit does not reach llms.txt straight away, because the plugin refreshes the file in the background. Saving a page selected for llms.txt schedules a WP-Cron event about 30 seconds later. Several saves in quick succession merge into one refresh. The editor therefore finishes saving at once, without waiting for the file to build. llms-full.txt works the same way, with its own background event.

To force the change right now, use the “Generate now” button on the module screen. The generator then rebuilds the content and refreshes the cache for all domains. The result of every run goes to the “Last 100 logs” list with a date and status. That is where you check whether the latest generation ended with OK or ERROR.

What should you do when WordPress cannot write the file on the server?

When WordPress cannot write the file on the server, the “Filesystem access” card in Global settings tells you in advance. The card checks robots.txt, llms.txt and llms-full.txt without any test writes. For each of them it shows whether the file can be created or overwritten in the WordPress root. The same status appears in a line below the form of each module. The messages tell several situations apart:

• WordPress cannot write files directly and would need FTP or SSH credentials,
• the root directory blocks writing, so a new file cannot be created,
• the file exists without write permission, so it cannot be overwritten,
• the robots.txt write goes through, but the read-back check does not match the content.

In each of these cases the “Download” button helps. It fetches exactly the content the plugin would write to disk. You then upload the file to the root directory yourself. Keep one thing in mind: that file will hide later panel changes until you replace it.

Why did the robots.txt write button disappear from the screen?

The robots.txt write button disappears when one file tree serves several sites with separate databases. A single file in the root would then reach all of those sites at once. The plugin therefore hides the “Save & write robots.txt file” button and refuses to write to disk. The status line explains that WordPress serves robots.txt dynamically. In such a setup llms.txt and llms-full.txt also run in dynamic mode only.

How do you check step by step which file crawlers actually see?

You can check which file crawlers actually see in a few steps, without touching your hosting. Start by comparing the preview with the live version. Any difference between them means something other than your settings is answering. Work through these steps in order:

1. open the robots.txt screen and compare the preview with the “Open live robots.txt” link,
2. check whether a warning about a physical file appears below the form,
3. make sure the “Serve generated robots.txt through WordPress” option is ticked,
4. in Global settings, review the “llms delivery” mode and the “Filesystem access” card,
5. on the llms.txt screen, click “Generate now” and look at the log.

A full description of the modules and screenshots is on the SEO/GEO Files Generator page. If you would rather have help with the setup, the BeeClear team can configure the files for your site. A few quick checks tell you whether crawlers read your rules or a forgotten copy.