llms.txt generator: a clear directory for AI readers
Generate a public llms.txt file, choose safe links and learn how to publish and maintain it without confusing it with robots.txt or a ranking guarantee.
Generate your llms.txt
# > ## Pages
What llms.txt is for
The llms.txt proposal describes a Markdown file at the root of a website that gives language-model readers a concise introduction and links to useful resources. It is intended to help a reader choose where to look next instead of navigating a large site blindly. The format is simple enough for a person to inspect, which makes it useful as an editorial directory as well as an experiment in AI-oriented discovery.
Support is not standardized across AI services. Publishing a file does not mean ChatGPT, Google or Perplexity will read it or use it to cite your business. It is not a prerequisite for Google AI features. Treat it as a maintained public map, not a hidden optimization hack. Your actual pages still need to be accessible, accurate and useful, and any private information must remain protected by real access controls.
Use a simple and valid structure
Start with one H1 containing the site name. Add a short blockquote summary if it helps explain the site. Optional paragraphs can clarify the product or the intended audience. Use H2 sections followed by Markdown link lists. Each list entry should include a descriptive link title and a short explanation of what the reader will find. Keep the structure flat and place an Optional section last if you want to distinguish lower-priority reading.
A good file is selective. It does not copy every paragraph from every page and it does not need to include every sitemap entry. Documentation, public pricing, service descriptions and substantial guides are useful candidates. Use stable canonical links instead of tracking URLs. Describe the actual page, not a keyword you hope to rank for. If a route redirects or has been removed, update the file along with the site's ordinary internal links.
Exclude private and operational routes
Do not include login, signup, password reset, OAuth callback or consent pages. Exclude account settings, customer dashboards, billing pages, profile editors and per-user reports. Administration, staff and internal operations pages are also unsuitable even when a route accidentally renders without authentication. A public directory should not advertise private entry points or encourage a crawler to explore operational surfaces.
The generator on this page accepts a site name, summary and a curated list of public links. It filters known private path families as a safeguard, but you remain responsible for reviewing what you publish. A custom path can still contain sensitive information even if its name looks harmless. Never paste access tokens, invitation links, passwords, customer details or unpublished documents into a public discovery file. Automated filtering cannot replace knowledge of your own website.
Generate, review and download
Enter the name that visitors actually see and write a plain-language summary of what the site provides. Add one public URL per line, optionally followed by a readable title. Choose links that a signed-out visitor is meant to land on. Preview the generated Markdown before downloading. Check the heading, spelling, summaries and destination pages. The file should be comprehensible to a person unfamiliar with your application.
The generator works in the browser and does not crawl your whole website or send the submitted page list to a content service. That avoids pretending an automatically discovered URL is safe merely because it exists. If you use a sitemap as a starting list, remove private and duplicate entries before pasting. The output is named llms.txt, but downloading it does not publish anything to your server.
Publish at the correct address
Place the reviewed file where your hosting setup serves public root files. For a conventional static deployment, that usually means adding it to the public assets folder so it is available at /llms.txt. A WordPress deployment may need a file upload or a hosting-level rule. Ask the person responsible for the site to verify the exact public URL. Do not assume a downloaded file has reached your website.
Fetch the result without logging in and check that it returns text rather than an HTML not-found page or a redirect to your dashboard. A successful HTTP status alone is not enough if your application serves its homepage for unknown filenames. Compare the actual body with the file you generated. If a caching layer is involved, verify the latest deployed version before investigating crawler behavior. Preserve the site's existing privacy and access settings.
Maintain it alongside SEO essentials
Keep your sitemap focused on real public canonical pages. Keep robots.txt aligned with intentional crawler access. Use page-specific titles and descriptions, visible headings and meaningful internal links. llms.txt does not replace any of these mechanisms, and it is not a way to undo a noindex directive or an authentication requirement. A useful directory points to content that already deserves to be read.
Update entries when a guide, feature or price materially changes. Do not add fresh dates unless you know when the page's content actually changed. Review the file after removing a route or reorganizing documentation. A short accurate file is preferable to a long stale one. Periodically test the linked pages as a signed-out visitor and remove entries whose access requirements have changed.
Auto-SEO publishes its own public directory and provides this free generator as a practical utility. For broader website work, start with the free page scan and inspect the report. If you create a project, use recommendations and approved actions to improve the pages themselves. No paid plan, generator or Markdown file can guarantee indexing, AI mentions or citations. The value here is clarity, safe public navigation and easier maintenance.
An approval-first website review
Check access
Open a public page without login and verify its initial HTML and crawler rules.
Review the evidence
Inspect the page findings and verify any proposed content with the business owner.
Apply an approved change
Use only supported publishing actions with the required permission.
Compare observations
Recheck comparable keywords or prompts, record dates and distinguish mentions from citations.
Questions and limitations
Does this guarantee rankings or AI citations?
No. Search engines and AI systems select results independently. Checks and improvements do not guarantee positions, traffic or citations.
How can I start?
Use the free public page scan without an account or card. Create a project when you want broader analysis and ongoing reports.
Sources and further reading
- Google: AI features and your website
- OpenAI: Publishers and developers FAQ
- Perplexity: Crawler documentation
- The llms.txt proposal
Written from Auto-SEO's documented workflows and primary platform guidance. Search data was checked in the US market; local demand varies. Provider documentation may change.