Best llms.txt Implementation Platforms: Where Your File Comes From

Your llms.txt file was almost certainly written by software you did not choose. On Shopify it is a platform route. On Mintlify it is a build artifact capped at 100,000 characters. On Framer it is whatever text file you last dragged into a settings panel. Every roundup of the best llms.txt implementation platforms currently ranking on that query was published by one of the products inside it, which is why a WordPress owner keeps reading advice designed for a documentation pipeline. This comparison groups sixteen named platforms by the mechanism that produces the file, records what each one emits with no configuration, names the ceiling each one hits, and gives the exact override path. Every behaviour below is traced to the vendor's own documentation and dated.

Key takeaways

• The best llms.txt implementation platforms divide into four mechanisms: a platform native route, a CMS plugin, a framework module, and a manual static upload. The sixteen platforms compared below fall into exactly one each.
• Mintlify caps automatically generated llms.txt files at 100,000 characters, then truncates and appends a note listing the number of omitted pages.
• Shopify made /agents.md its primary storefront discovery file on 20 May 2026 and now redirects /llms.txt and /llms-full.txt to it.
• Ahrefs checked 137,210 domains in May 2026 and found that 97% of the llms.txt files it located received zero requests that month.
• Gary Illyes of Google told a Search Central Deep Dive event that "Google doesn't support LLMs.txt and isn't planning to", as reported by Search Engine Land in July 2025.

Why Every llms.txt Roundup Ranks Its Own Author First

llms.txt is a plain Markdown file at a domain root that lists a site's most important pages with a one line description each, so a language model can navigate the site without crawling every URL. Jeremy Howard, co-founder of Answer.AI, published the proposal on 3 September 2024 alongside an optional companion file, llms-full.txt, which carries the full text of the site in a single document.

Search the phrase and count the authors. Mintlify's platform roundup, Fern's January 2026 comparison and Context.dev's tool list all rank on the first page, and each one is published by a product that appears in its own ranking. AIOSEO's guide to WordPress generators puts AIOSEO at number one. The pattern is structural rather than dishonest: a vendor picks the evaluation criteria on which it already performs well.

The skew has a direction. Most of the highest ranking comparisons treat documentation as the default site shape, because documentation platforms solved this first and have the strongest incentive to publish about it. A Shopify merchant, a Drupal editor, or an agency running eleven Webflow sites gets nothing applicable from any of them, even though all three arrive on the query looking for the same thing: the best llms.txt implementation platforms for the stack they already run.

This article was written by a studio that sells a paid llms.txt generator, so take the bias declaration first: nothing below recommends a generator, including ours. The best llms.txt implementation platforms are ordered here by mechanism, never by vendor. The comparison covers where the file is produced, not what belongs inside it. The editorial question of what deserves a line in the file is handled separately in generating an llms.txt that earns its place at your domain root.

One set of numbers should frame every choice that follows. Ahrefs examined all 137,210 domains in its Web Analytics product that received traffic in May 2026, found that 28% of them publish an llms.txt file, and measured that 97% of those files received zero requests during the month. Among the 3% that were fetched, 96% of requests came from bots, with named AI tools accounting for 19.5% and SEO audit tools for 21.7%. Evil Martians reached a similar conclusion from a different angle, tracking roughly 268,000 agent requests between early May and early July 2026 and logging about 660 direct fetches of /llms.txt, of which around 37 came from named AI assistants. Choose whichever of the best llms.txt implementation platforms costs you the least ongoing effort, because the evidence does not support spending much.

The Best llms.txt Implementation Platforms, Grouped by Four Mechanisms

Four mechanisms produce the file, and every one of the best llms.txt implementation platforms uses exactly one of them. A platform native route means the host generates and serves it as part of the product, with no code from you. A CMS plugin means a third party module reads your content database on a schedule. A framework module means the file is a build output of your static site generator or meta framework. A static upload means you produce the text somewhere else and drop it into a file host. The mechanism determines who owns the refresh cycle, which is the only property that still matters six months after launch. The sixteen rows below cover the best llms.txt implementation platforms by installed base and by quality of published documentation, grouped by that mechanism rather than ranked.

Read that table by row, not by rank. There is no single winner among the best llms.txt implementation platforms, because the mechanism is decided by where your content already lives. Migrating a storefront to a documentation host to obtain a better generated file costs more than the file is worth.

Platform Native: Shopify and What agents.md Changed

Shopify is the only platform in this comparison that serves three discovery files as native routes with no configuration. It generates /llms.txt, /llms-full.txt and /agents.md at the platform level as native storefront routes, with no app installation. On 20 May 2026 the platform promoted /agents.md to primary discovery file and began redirecting the two llms files to it, a change surfaced in the Shopify developer community rather than in the changelog. Merchants who had built monitoring against /llms.txt found out by watching their checks follow a redirect.

agents.md is a Markdown file addressed to autonomous agents rather than to answer engines: it describes what actions a site supports, not merely which pages exist. For a storefront the distinction is real, because an agent buying a product needs the cart and checkout surface, not a page index.

The override path is theme templates. Create templates/agents.md.liquid, templates/llms.txt.liquid or templates/llms-full.txt.liquid in your theme and Shopify serves your version instead of the generated one. The template is Liquid, so the full storefront object model is available: collections, products, metafields and policy pages. Nothing else in this comparison offers that much control from inside the platform. The concrete steps live in overriding the Shopify agents.md and llms.txt defaults.

The ceiling is judgement. A Liquid template renders against live storefront data, so it cannot summarise. Deciding which twelve of your four hundred collections deserve a listing has to be encoded as a metafield or hardcoded into the template by a person.

Documentation Hosts: Mintlify, GitBook and Fern, the Best llms.txt Implementation Platforms for Docs

Documentation platforms shipped this first and document their behaviour most precisely. Mintlify hosts llms.txt and llms-full.txt at the project root and also at /.well-known/llms.txt and /.well-known/llms-full.txt, and states in its own reference that the files are "always up to date and requires zero maintenance".

Mintlify is the only platform in this comparison that publishes a hard character ceiling, and three limits sit inside its zero maintenance promise. Automatically generated llms.txt files cannot exceed 100,000 characters; past that, Mintlify truncates the file and appends a note listing the number of omitted pages. Page descriptions truncate at 300 characters and at the first line break, whichever arrives first. Pages are listed alphabetically in the order they appear in the repository, starting from the root directory, so the file's structure mirrors your folder names rather than your reader's priorities.

Authentication changes the output entirely. On a fully authenticated Mintlify site both files require authentication, which means no agent reads either one. On a partially authenticated site both stay public but list only public pages, excluding anything restricted to a user group.

The override is a file commit: adding a custom llms.txt overrides the generated file of the same name, and that custom file must carry a site title as an H1 heading. GitBook added llms.txt support in January 2025 and followed with llms-full.txt and per page Markdown in June 2025, alongside an MCP server per space. Fern regenerates token optimised llms.txt and llms-full.txt on every documentation change and serves Markdown instead of HTML when it detects LLM traffic, which cuts the tokens an agent spends reading a reference page.

For a documentation estate, the deciding number among the best llms.txt implementation platforms is the ceiling: a four hundred page API reference will cross 100,000 characters and lose pages, and only Mintlify tells you how many.

CMS Plugins: WordPress, Drupal and What They Inherit From Your SEO Plugin

WordPress and Drupal are the two content management entries among the best llms.txt implementation platforms, and they solve the same problem in opposite directions. The Website LLMs.txt plugin has passed 40,000 active installations and sits at version 8.5.3, requiring WordPress 5.8 or later and tested to 6.9.5, according to its WordPress.org listing. It writes a curated llms.txt of public URLs plus an optional llms-full.txt, and rebuilds immediately on content change, daily, or weekly, with a manual Generate Now button in settings.

Its most valuable behaviour is inherited rather than configured. The plugin reads Yoast SEO, Rank Math, SEOPress and All in One SEO, and automatically excludes anything those plugins marked noindex or nofollow. Your existing search hygiene becomes your llms.txt inclusion policy at zero extra effort, which is the strongest single argument for the CMS plugin mechanism anywhere in this comparison.

Control runs out at ordering and at review. You can include or exclude post types and optionally add taxonomies, but the file ships the moment it regenerates. There is no staging copy and no diff against the previous version, so a bulk edit across 300 posts rewrites your agent facing file with nobody reading it. Teams that care put the plugin on the weekly schedule and read the output before each rebuild. The install path is in adding llms.txt to WordPress without touching your theme.

Drupal takes the opposite position, and it is the most deployment aware of the best llms.txt implementation platforms. The /llms.txt module on Drupal.org, at version 1.0.7 as of 28 April 2026 with support for Drupal 10.3 and 11.1, splits the file across two stores: structural sections held as exported Drupal configuration at /admin/content/llms-txt, and environment specific entries held as content entities at /admin/content/llms-txt/sections. Structure travels from development to production through your normal configuration workflow while per environment content stays where it was written. The module's documentation adds a server note that catches people on first deploy: under Nginx, add a location block for /llms.txt with try_files $uri @drupal;, or the request never reaches Drupal at all.

Framework Modules: Nuxt, Astro Starlight, Docusaurus and Hugo

Framework modules are the most reviewable of the best llms.txt implementation platforms, because the file lands in a pull request. A framework module turns the file into a build artifact, versioned in the same repository as the content it describes. The nuxt-llms module reached version 0.2.0 with 97,700 downloads and 165 GitHub stars on the Nuxt module registry. Two of its defaults catch people. The domain option is mandatory, because the specification expects absolute URLs. And llms-full.txt is not generated at all until both full.title and full.description are set, so a Nuxt site that appears to have shipped half the pair has shipped exactly what the module does out of the box.

The module exposes two runtime hooks, llms:generate on every /llms.txt request and llms:generate:llms-full on every /llms-full.txt request. Both run per request rather than at build time, which is the cleanest editorial control in the framework category: you can reorder sections, drop pages, or inject a paragraph of positioning without touching a single content file. Nuxt Content 3.2.0 and later feeds it directly.

starlight-llms-txt emits three files for Astro Starlight rather than two, adding llms-small.txt, a filtered version sized for smaller context windows, and takes a project name, exclude patterns and custom element selectors as configuration. docusaurus-plugin-llms does the equivalent job for Docusaurus, emitting llms.txt, llms-full.txt and per page Markdown with batch processing for large sites.

Hugo and Jekyll have no first party module and do not need one. Hugo produces the file through a custom output format declared in the site configuration with isPlainText enabled, which tells Hugo to render through Go's text/template package so that angle brackets and quotes in your Markdown survive unescaped; you then write a template that groups pages by section. Jekyll works the same way, through a source file carrying layout: null front matter and rendered by Liquid at build time, which is the same technique that produces its sitemap.

Route Handlers and Build Steps: Next.js With a CMS Behind It

Next.js sits apart from the rest of the best llms.txt implementation platforms because it ships file conventions for robots.ts and sitemap.ts and none for llms.txt. An open discussion on the vercel/next.js repository asks for an llms.js convention mirroring the other two. Until that lands, the file is a route handler you write.

The decision is one line long. Create app/llms.txt/route.ts with a GET handler returning plain text. Add export const dynamic = 'force-static' and the file is generated once at build, which is right for a marketing site whose page set only changes on deploy. Leave the handler dynamic and it runs per request, which is right when a headless CMS publishes independently of your deploy pipeline.

The trap is the middle case: a statically generated route on a site whose CMS publishes daily. The file then describes the site as it stood at the last deploy, and nothing surfaces the drift. Either move to a dynamic handler with an explicit cache header, or trigger a rebuild on publish. The working implementation is in a Next.js route handler that builds llms.txt from your CMS.

Static Upload: Framer, Webflow, Wix and Squarespace

Three of the four static upload entries among the best llms.txt implementation platforms gate file hosting behind a paid plan tier, which is the first thing to check before promising a client the file exists.

Framer hosts static files on Pro, Scale and Enterprise plans only; Basic plans do not support it. You upload through Domains, then Files in the dashboard, then publish, and the file resolves at your domain root. Framer accepts TXT and JSON, so a structured machine readable variant is possible alongside the plain text one.

Webflow gates the upload to CMS and Business site plans. The control sits in Site Settings under the Publishing tab. Webflow serves the file on the published custom domain, so verification happens after publishing rather than on staging. Changing the file means re-uploading it and republishing the site.

Wix is the exception in this group: it generates the file automatically from public content and lets you edit it in the dashboard under SEO and GEO, then Tools and settings, then Go to llms.txt. Editing carries a consequence Wix states plainly, that once you edit the file it stops updating automatically so your changes survive. A reset returns it to the automatic version. A privacy toggle removes it from the site entirely. Wix also publishes a separate help article titled "SEO Request: Making LLMs.txt Available for All Sites", which is how a platform signals a feature that has not reached every plan.

Squarespace exposes no document root to site owners. Uploaded files land on its CDN under generated asset paths, so /llms.txt is not a file but a URL mapping: upload the text file to the asset manager, copy its CDN URL, then add a 301 redirect from /llms.txt to that path under URL Mappings. That route is documented across Squarespace's own community forum rather than in a help article, which is itself a signal about how supported it is. The redirect resolves correctly and it is also the most fragile arrangement in this comparison, because the asset URL changes whenever the file is replaced rather than overwritten in place.

This is where manual upload breaks. A static file describes a content set frozen at upload time. Add twelve products, rename a collection or retire a landing page and nothing updates. Among the best llms.txt implementation platforms, the static upload group is the only one where drift is guaranteed rather than merely possible, so budget a recurring calendar reminder alongside the upload. The four paths are compared in the no-code upload path for llms.txt.

What Each Mechanism Silently Decides for You

Every one of the best llms.txt implementation platforms takes seven decisions out of your hands. Know them before choosing. Each is a point where a generated file diverges from a file a person would have written.

1. Ordering. Mintlify lists pages alphabetically from the repository root. Plugins usually order by publication date or menu order. Neither matches the order in which you would introduce your business to a stranger.
2. Inclusion. WordPress plugins inherit noindex from your SEO plugin. Wix includes only public content. A Shopify template decides for itself. None of them asks whether a page deserves a model's attention, only whether it is public.
3. Description text. Mintlify truncates each description at 300 characters and at the first line break. Plugins reuse the meta description written to win a click from a human, which is a different job with a different reader.
4. Truncation. Mintlify stops at 100,000 characters and reports how many pages it dropped. Most other mechanisms publish no ceiling at all, which makes their failure silent rather than announced.
5. Refresh trigger. Platform native and framework routes rebuild on deploy or on request. Plugins rebuild on a schedule. Static uploads rebuild when a human remembers, and the Ahrefs data suggests most do not.
6. File location. Mintlify also serves /.well-known/llms.txt. Shopify redirects /llms.txt to /agents.md. Squarespace serves a redirect to a CDN path. An agent checking exactly one path finds nothing on at least one of these.
7. Deploy coupling. This is the sharpest split across the best llms.txt implementation platforms: a framework module means the file cannot change without a deploy. That is a feature for review discipline and a defect for a marketing team that publishes on a Friday afternoon.

The Decision Table: Site Shape to Mechanism

Map the site shape to the mechanism, then read the platform column. This is the shortest route through the best llms.txt implementation platforms for a technical buyer with several stacks to serve.

Two rows deserve a caveat. Wix is the only no-code host that regenerates for you, which makes it the least demanding member of the upload group by a wide margin. And the storefront row assumes you accept the 20 May 2026 redirect: if a partner integration or an uptime check expects /llms.txt to return content rather than a redirect, the template override stops being optional. Across the best llms.txt implementation platforms, the ones that ask least of you are the ones already holding your content.

Two Cases Where You Should Write the File by Hand

Automation is the right default everywhere else in this comparison. Two situations invert it.

A site under roughly forty pages. Below that size the file is not an index, it is a positioning document. A generated list of forty URLs with their meta descriptions tells a model that your site exists. Forty lines written by a person tell it what you do, who you serve, and which three pages answer the questions you actually get asked. No generator among the best llms.txt implementation platforms produces that, because none of them knows your business. The work is one afternoon and the file changes twice a year.

A multi platform estate. A company running WordPress for the blog, Shopify for the store and Mintlify for the docs ends up with three separately generated files across three hosts, each describing a fragment of one business and none of them referencing the other two. No mechanism in this comparison reconciles them, because every mechanism sees only its own content database. Writing one file by hand for the primary domain and linking the other two from it is the only arrangement that presents the company as a single entity, and that problem is worked through in managing llms.txt across multiple client sites.

FAQ: Choosing Among the Best llms.txt Implementation Platforms

Which of the best llms.txt implementation platforms requires the least maintenance?

Platform native builds. Mintlify, GitBook and Fern regenerate on every documentation change with no configuration, and Mintlify states the files require zero maintenance. Wix regenerates automatically for no-code sites. The trade is control: you accept alphabetical ordering, inherited descriptions and, on Mintlify, a 100,000 character ceiling that truncates the file and reports the omitted page count after the fact.

Does Shopify still serve /llms.txt?

It returns a redirect, and among the best llms.txt implementation platforms Shopify is the one whose documented path moved under live traffic. Since 20 May 2026 Shopify has treated /agents.md as the primary storefront discovery file and redirects both /llms.txt and /llms-full.txt to it. The three Liquid templates remain the documented override path, so a merchant who needs /llms.txt to return content can start there, then verify the live behaviour on the storefront before promising it to a partner integration.

Why does my Nuxt site have llms.txt but no llms-full.txt?

Because nuxt-llms does not generate it by default. The module documentation states that the /llms-full.txt route is created only once both full.title and full.description are set in the module configuration. Adding those two values and redeploying produces the file. The domain option is separately mandatory, since the llms.txt specification expects absolute URLs rather than site relative paths.

Do WordPress llms.txt plugins respect noindex?

The Website LLMs.txt plugin does. It reads Yoast SEO, Rank Math, SEOPress and All in One SEO and automatically excludes content those plugins marked noindex or nofollow, per its WordPress.org listing. That inheritance is the mechanism's main advantage and its main limitation at the same time: inclusion is decided by your search settings, not by whether a page is useful to a language model.

Is llms.txt worth implementing if AI systems ignore it?

Measure the cost rather than the promise. Ahrefs found that 97% of the llms.txt files it located across 137,210 domains received zero requests in May 2026, and Gary Illyes of Google said at a Search Central Deep Dive event that Google does not support the file and does not plan to, as reported by Search Engine Land in July 2025. A mechanism costing one configuration change stays defensible at that hit rate. A quarterly manual regeneration ritual does not.

Which of the best llms.txt implementation platforms handle a site with thousands of pages?

A framework module or a route handler, because both filter before the file is written, which separates them from every other entry among the best llms.txt implementation platforms. Mintlify truncates at 100,000 characters and reports the omitted page count. Plugin based generation on a large WordPress install produces an unfiltered dump with no token budget. docusaurus-plugin-llms batches large sites explicitly, and a Next.js route handler can select the top pages by traffic before rendering a single line.

How La Boétie Ships the Agent-Facing Layer

La Boétie builds and runs the agent facing layer across client estates on all four mechanisms above, on most of the best llms.txt implementation platforms named in this comparison, and sells the one piece no platform generates for you.

The file itself. The studio's llms.txt Generator is a self serve tool at $25 or 20 € a run, with no account and no call. It crawls the site, selects the pages worth listing and writes the descriptions, which is the editorial step every platform native generator skips. Run the llms.txt Generator on a domain and compare its output line by line against whatever your platform emits today.

Multi platform estates. A flexible team of five to six engineers, multilingual and spread across timezones, has shipped client sites on WordPress, Shopify, Next.js and static hosts, including france-epargne.fr, assurecompare.fr and llb-auction.com. Reconciling three separately generated files into one coherent account of a business is engineering work, not a settings change.

Ownership. Every mechanism in this comparison is a dependency on somebody else's default. The studio takes its name from Étienne de La Boétie, whose 1548 argument against voluntary servitude is the thesis applied here: the file describing your business belongs to your business. Clients keep ownership of everything built.

Conclusion

The file's origin is settled before you write a line of it. The best llms.txt implementation platforms are not ranked against each other, they are matched to a stack. Sixteen platforms, four mechanisms, and in every case the mechanism belongs to wherever your content already lives. Shopify and Wix generate and let you override. Mintlify, GitBook and Fern generate and truncate. WordPress and Drupal inherit their inclusion rules from tools you configured for a different purpose entirely. Nuxt, Astro Starlight, Docusaurus, Hugo and Jekyll turn the file into a build artifact you can review in a pull request. Framer, Webflow and Squarespace hand you an upload box and a plan tier.

Pick by row in the decision table, not by rank in somebody's roundup. The best llms.txt implementation platforms for your site are the ones already holding your content, and the only choice worth deliberating is whether the descriptions inside the file were written by a person who knows what the business does. Everything else is a default you inherited on the day you picked a host.

Sources

Further reading

• Generating an llms.txt that earns its place at your domain root
• Adding llms.txt to WordPress without touching your theme
• Overriding the Shopify agents.md and llms.txt defaults
• A Next.js route handler that builds llms.txt from your CMS
• The no-code upload path for llms.txt
• Documentation platforms generate the file, then you edit it
• Managing llms.txt across multiple client sites

External sources

• llms.txt : Mintlify Documentation, 2026
• llms.txt and agents.md : Shopify Developer Community, 2026
• Website LLMs.txt : WordPress Plugin Directory, 2026
• nuxt-llms : Nuxt Modules, 2026
• The /llms.txt module : Drupal.org, 2026
• llms.txt in Framer : Framer Help Center, 2026
• Upload an llms.txt file to your site : Webflow Help Center, 2026
• Understanding your site's LLMs.txt file : Wix Help Center, 2026
• We analyzed 137K sites: 97% of llms.txt files never get read : Ahrefs, 2026
• Which AI actually reads your site : Evil Martians, 2026
• The /llms.txt file, a proposal to provide information to help LLMs use websites : Answer.AI, 2024
• Google says normal SEO works for ranking in AI Overviews and llms.txt will not be used : Search Engine Land, 2025
• starlight-llms-txt : GitHub, 2026
• LLM ready docs : GitBook Documentation, 2026
• llms.txt : Fern Documentation, 2026
• LLMS.txt files on Squarespace sites : Squarespace Forum, 2026