# How to Use an llms.txt Generator (and When Not To)
> An llms.txt generator drafts the file in seconds, but 97% of llms.txt files never get read. Here is how to build one that earns its place.

Canonical URL: https://cite.solutions/blog/how-to-use-an-llms-txt-generator
Source: Cite Solutions (cite.solutions)
Published: 2026-07-11
---

[Technical Guides](/category/technical-guides)10 min read

# How to Use an llms.txt Generator (and When Not To)

[Subia PeerzadaFounder, Cite Solutions · July 11, 2026](https://www.linkedin.com/in/subia-peerzada-75025764/)

Key takeaways

## using an llms.txt generator

A generator drafts the file in seconds, but the version that earns its place is hand-curated: a short list of canonical, answer-first pages, matched to your real site structure.

1. 01A generator flattens your sitemap into the llms.txt format. That is a draft, not a citation lever. A study of 137,210 domains found 97% of llms.txt files got zero requests in May 2026.
2. 02Google confirmed its Search does not read llms.txt. Build one for agent and documentation discovery, not for a ranking bump that is not there yet.
3. 03The value is the curation: a short file pointing only at canonical, answer-first pages, matched to how your site is actually linked.

An `llms.txt generator` will hand you a finished file in about thirty seconds. Paste your domain, it crawls your sitemap, and out comes a formatted `llms.txt` ready to drop at the root of your site. The pitch is that this helps AI systems find and cite your best pages.

Here is the honest version. The generator does the easy 20% of the job and skips the part that matters. Below is what these tools actually produce, why most of their output gets ignored, and how to turn a generated draft into a file worth having.

## An llms.txt generator gives you a draft, not a ranking

An llms.txt generator scaffolds the file by pulling your sitemap into the `llms.txt` format: an H1 with your site name, a summary, and lists of links. Useful as a starting point. It will not get you cited. A study of 137,210 domains by [Ahrefs](https://ahrefs.com/blog/llmstxt-study/) found 97% of llms.txt files got zero requests in May 2026, and Google has confirmed its Search does not read the file.

So the question is not "which llms.txt generator should I use." The question is what the file is for, and whether the auto-generated version does that job.

The short answer: a generator is fine for the format and useless for the judgment. The format takes thirty seconds. The judgment, which pages belong in the file and which do not, is the entire point.

If you want the background on what the file is and whether your site needs one at all, we covered that in [llms.txt: what it is and whether your site needs it](/blog/llms-txt-what-it-is-and-why-your-site-needs-one). This guide is narrower: how to use a generator without shipping the slop it produces by default.

## Why a generated llms.txt file usually gets ignored

The problem is not the generator's formatting. The format is trivial and the tools get it right. The problem is that a machine-built file inherits everything wrong with your sitemap and adds nothing a model was missing.

A generator answers "what URLs exist." AI systems were never confused about that.

### Reason #1: 97% of llms.txt files got zero requests last month

The Ahrefs study tracked 137,210 domains, 28% of which published an llms.txt file. In May 2026, 97% of those files received zero requests. Of the tiny fraction that got any traffic, most fetches came from SEO audit tools and general crawlers, not AI systems. We dug into that crawler behavior in [do AI crawlers read llms.txt](/blog/do-ai-crawlers-read-llms-txt). A generated file that nothing reads earns nothing.

### Reason #2: Google confirmed Search does not read your llms.txt

Google's own AI optimization guidance states you do not need to create machine-readable AI files, markup, or Markdown to appear in Google Search or its AI features, because Search does not use them. Gary Illyes said at Google Search Central Live that Google does not support llms.txt and has no plans to. A generator cannot manufacture a reader that does not exist.

### Reason #3: No major AI vendor has committed to the file as a signal

John Mueller compared llms.txt to the old keywords meta tag: something a site owner claims about their own site, which is exactly why engines learned to ignore it. As he [put it](https://www.seroundtable.com/google-does-not-endorse-llms-txt-40789.html), you can look at your server logs and see the AI services do not even check for the file. Why trust a self-report when you can read the page directly.

### Reason #4: A generator ships your sitemap, not a curated map

The whole idea of llms.txt is a short, opinionated list of what matters. A generator does the opposite. It flattens every URL it can find, including thin pages, tag archives, and interface routes. The output is a sitemap with different punctuation. That defeats the one thing the file was supposed to do.

### Reason #5: The file cannot rescue pages that cannot earn a citation

Pointing an AI system at a weak page does not make it citable. If the page buries its answer, cites no sources, and reads like a brochure, listing it in `llms.txt` changes nothing. Citations come from [passage-level structure](/blog/passages-beat-pages-how-to-structure-content-for-ai-citation), not from a directory file that names the page.

Here is the split in plain terms.

**A generator asks:**

* •Which URLs are in the sitemap?
* •How do I format them as llms.txt?
* •How fast can I output a file?

**A file worth reading asks:**

* •Which handful of pages would I actually want a model to quote?
* •Does the summary say what we cover and who it is for?
* •Do these URLs match how the site is really linked and navigated?

The first file is an export. The second is an editorial decision.

llms.txt generator reality check

### A generator makes the file. It does not make it worth reading.

97%

of llms.txt files got zero requests in May 2026 (Ahrefs, 137,210 domains)

#### What a generator ships

Draft in 30 seconds

* •Your full sitemap flattened into the llms.txt format
* •Every URL, including thin pages and interface routes
* •A summary line pulled from your meta description
* •No judgment about which pages can actually earn a citation

#### What a file worth reading has

Curated by hand

* •A short list of canonical, answer-first pages a model could quote
* •A summary that says what you cover and who it is for, in plain words
* •URLs that match how your site is actually linked and navigated
* •A date stamp and a review cadence, treated as living infrastructure

## When an llms.txt generator is still worth your time

None of this means skip the file. It means be honest about why you are shipping it. There are real reasons, and none of them is a near-term citation bump.

A generated draft is a fine starting point when you plan to edit it down.

Use one when any of these is true:

* •You run documentation or a developer product, where AI agents and coding assistants genuinely benefit from a compact, linked map of your resources.
* •You want the curation exercise. Deciding which ten pages belong in the file forces a useful conversation about which pages are actually your best.
* •You want cheap future-proofing. If vendors do adopt the file, the cost of having a clean one now is close to zero.
* •You already track AI visibility and treat llms.txt as one small layer, not the strategy.

Skip the anxiety when your site is five marketing pages with obvious navigation. Your citation problem is not a missing file. It is not having enough pages a model would want to quote yet.

The table below shows the gap between what a generator hands you and what actually belongs in the file.

| Element     | Generator default            | What earns its place                                  |
| ----------- | ---------------------------- | ----------------------------------------------------- |
| Page list   | Full sitemap, every URL      | 5 to 15 canonical, answer-first pages                 |
| Summary     | Pulled from meta description | Written in plain words: what you cover, who it is for |
| URL choice  | Whatever the crawler found   | Pages that match your real internal linking           |
| Maintenance | Static, never revisited      | Date-stamped, reviewed when priorities change         |
| Purpose     | Claim coverage               | Point agents at your strongest resources              |

### Not sure which pages belong in your llms.txt?

We audit the pages AI systems actually retrieve, fix weak answer structure, and set up practical layers like llms.txt where they help. You get a file built around the pages worth citing, not a sitemap dump.

[Book a Technical GEO Review](/contact)

## How to build an llms.txt file that earns its place

You can start from a generator. Just do not stop there. Treat the auto-generated output as a rough draft and spend fifteen minutes turning it into something curated. Work through these five steps in order.

A short file pointing at your best pages beats a long file pointing at all of them.

## Step 1: Start from the generator, then cut it in half

Run a generator to get the format and the raw URL list, then delete most of it. Keep only pages you would genuinely want an AI system to quote or an agent to open first. For most sites that is five to fifteen URLs, not five hundred. If you cannot say why a page is on the list, it is not on the list.

## Step 2: Point only at canonical, answer-first pages

Every URL in the file should be the canonical version and should lead with a direct answer, not a hero image and a sign-up form. A model that follows the link should hit a clean passage it can lift. Pages that bury the answer or duplicate another URL waste the slot and teach the file to point at the wrong version.

## Step 3: Write a summary a model can actually use

Replace the auto-pulled meta description with two or three plain sentences: what your company does, what the site covers, and who it is for. This is the one block the spec marks as important, and it is the part generators handle worst. Skip the adjectives. State the facts a model would need to place you correctly.

## Step 4: Match the file to your real site structure

If the file says a page matters, that page should also be easy to reach through your navigation and internal links. The file should reinforce your information architecture, not contradict it. Group the links under clear H2 sections, such as core pages, documentation, and research, so the map reflects how your site is actually organized.

## Step 5: Ship it, then watch your server logs

Place the file at `https://cite.solutions/llms.txt` for your domain, then check your server logs over the following weeks for requests to it. This is the only honest measure of whether anything reads it. Given the Ahrefs data, expect little at first. Revisit the file when you publish major new work or replace an old guide with a stronger one.

Here is what a curated file looks like, cut down from what a generator would have produced:

```txt
# Cite Solutions

Cite Solutions is a managed AI visibility service focused on GEO and AEO.
We help B2B brands understand and improve how they appear in ChatGPT,
Perplexity, Gemini, and other AI search surfaces.

## Core pages
- https://cite.solutions/geo-services
- https://cite.solutions/ai-visibility-audit
- https://cite.solutions/contact

## Key educational resources
- https://cite.solutions/blog/ai-citations-how-they-work
- https://cite.solutions/blog/passages-beat-pages-how-to-structure-content-for-ai-citation
- https://cite.solutions/blog/how-to-optimize-for-chatgpt-search

## Research and methodology
- https://cite.solutions/ai-search-statistics
- https://cite.solutions/blog/half-life-of-ai-citations

```

That is fifteen minutes of editing on top of a thirty-second generation. The editing is the part that matters.

## How to know if it is doing anything

Do not measure llms.txt by citations. The file is too far upstream, and the data says almost nothing reads it yet. Measure the two things it can plausibly affect and ignore the rest.

The only real signal is your logs. If AI bots never request the file, it is not the file doing the work.

Check two things. First, grep your server logs for requests to `/llms.txt` and note which user agents fetch it, if any. Second, watch whether the pages you listed get cited in AI answers at all, which tells you whether those pages deserve their spot regardless of the file. That second signal comes down to [how AI platforms choose which sources to cite](/blog/how-ai-platforms-choose-which-sources-to-cite), and the file has no vote in it. That second signal is the one that matters, and it comes from the page, not the directory. Our [first-party AI search statistics](/ai-search-statistics), drawn from 34,000-plus AI answers, found ChatGPT cites a source in 87% of its responses. Those citations are earned by the pages, and a clean llms.txt only helps a model find pages that were already worth citing.

If you would rather not run that loop yourself, a [managed AI visibility team](/geo-services) can build the file around the pages worth citing and track whether anything actually reads it.

## FAQ

### Do I need an llms.txt generator?

Only to save time on formatting. A generator produces the file structure in seconds, which is genuinely convenient, but the useful part is deciding which pages belong in the file, and no tool can do that for you. Use a generator for the draft, then cut it down to the handful of canonical, answer-first pages you would want a model to quote.

### What should an llms.txt file contain?

Per the [llmstxt.org spec](https://llmstxt.org/), the only required element is an H1 with your site or project name, followed by a short summary in a blockquote. After that, add H2 sections listing your most important pages as Markdown links, with a short note on each. Keep it curated. A good file names your best pages, not every page.

### Does Google use llms.txt?

No. Google confirmed in its AI optimization guidance, updated in 2026, that Search does not use llms.txt or any other AI-specific file, and Gary Illyes said Google has no plans to support it. Build the file for AI agents and documentation discovery if those matter to you, not for a Google ranking effect, because there is not one.

### Is llms.txt worth it in 2026?

For documentation-heavy and developer-facing sites, it is a low-cost layer worth having. For a small marketing site, it is optional and not urgent. The [Ahrefs study of 137,210 domains](https://ahrefs.com/blog/llmstxt-study/) found 97% of llms.txt files got zero requests, so treat it as cheap future-proofing and a curation exercise, not a visibility lever. Fix your page-level citation readiness first.

### What is the difference between llms.txt and llms-full.txt?

`llms.txt` is the compact map: a summary and a curated list of links. `llms-full.txt` (sometimes generated as `llms-ctx-full.txt`) inlines the full content of those pages into one large file so a model can read everything without following links. The full version is mainly useful for documentation sites feeding a specific model context. Most sites only need the compact `llms.txt`.

## The bottom line

An llms.txt generator is a formatter, not a strategy. It hands you a draft in thirty seconds by flattening your sitemap, and if you ship that draft unedited, you have published a sitemap with different punctuation that, per the data, almost nothing will read.

The version worth having is the one you curate: a short list of canonical, answer-first pages, a plain-language summary, and URLs that match how your site is actually built. Use the generator for the thirty seconds it saves. Spend the fifteen minutes it does not.

Because in AI search, the file only helps a model find pages that already deserve the citation. Earn that first.

### Build the technical layer of AI visibility on the right foundation

Cite Solutions audits the pages AI systems retrieve, fixes weak answer structure, and implements practical layers like llms.txt where they earn their keep. We start with the pages worth citing, then wire the file around them.

[Book a Discovery Call](/contact)

Tags

[llms.txt](/tag/llms-txt)[GEO](/tag/geo)[AEO](/tag/aeo)[technical SEO](/tag/technical-seo)[how to](/tag/how-to)[AI visibility](/tag/ai-visibility)[answer engine optimization](/tag/answer-engine-optimization)

## Continue the brief

[01Technical Guidesllms.txt: What It Is, What It Does, and Whether Your Site Actually Needs Itllms.txt is getting a lot of attention in GEO and AEO circles, but most of the commentary swings between hype and dismissal. Here's what the file actually does, what it cannot do, and when it is worth implementing.Apr 7, 2026Read→](/blog/llms-txt-what-it-is-and-why-your-site-needs-one)[02Technical GuidesDo AI Crawlers Actually Read llms.txt?Across 500M+ AI bot visits, only 408 fetched llms.txt. Here is what AI crawlers really do with the file, and whether it is worth publishing.Jun 7, 2026Read→](/blog/do-ai-crawlers-read-llms-txt)[03Technical GuidesLLM Optimization: How to Get Cited by AILLM optimization is how you get your brand cited inside AI answers. Here are the five levers that move citations and the steps to run them.Jun 16, 2026Read→](/blog/llm-optimization-how-to-get-cited)

[FrameworkLearn the CITE framework behind our GEO and AEO workSee how Comprehend, Influence, Track, and Evolve turn AI visibility into an operating system.](/framework)[ServicesExplore our managed GEO services and AEO execution modelAudit, prompt discovery, content execution, and ongoing monitoring tied to AI search outcomes.](/services)[AuditStart with an AI visibility audit before executionUnderstand prompt coverage, recommendation gaps, source mix, and where competitors are winning.](/ai-visibility-audit)

On this page

On this page

## Work with us on this

[AEO ServicesAnswer engine optimization: be the answer AI quotes.Explore→](/aeo-services)[GEO AgencyManaged generative engine optimization for B2B brands.Explore→](/geo-agency)[AEO AgencyAn agency built for answer engine optimization.Explore→](/answer-engine-optimization-agency)

## Ready to become the answer AI gives?

Book a 30-minute discovery call. We'll show you what AI says about your brand today. No pitch. Just data.

[Book a Discovery Call](/contact)