How to Implement llms.txt: Full Guide | Opere18
← Blog/AI Search

How to Implement llms.txt: Full Guide

August 10, 2026·6 min read·Ratish Rajendran

You know what llms.txt is and why it matters. This is the implementation walkthrough: the exact file structure, what to include and leave out, where to host it, and how to check it is actually being read, so you can go from zero to a working file in under an hour.

Before you start

If you have not read what llms.txt is and why it matters, the short version: it is a markdown file at yoursite.com/llms.txt that curates which pages an AI model should treat as canonical. This post is the how, not the why. You need a text editor, access to your site root, and about thirty minutes.

Step 1: Set up the file structure

llms.txt follows a simple markdown convention: an H1 with your brand name, a one-line blockquote description, then H2 sections grouping links with short descriptions. Keep the description identical to the one you use everywhere else, this file is part of your brand entity, not a separate voice.

SectionWhat goes in it
H1Your exact brand name
BlockquoteOne-line description, same wording as your homepage
## Core pagesHomepage, services, pricing, about
## Documentation / guidesYour most valuable how-to and pillar content
## OptionalSecondary pages worth including but not essential

Step 2: Choose which pages make the cut

This is the actual work, curation, not writing. Include your homepage, core service pages, pricing, and the handful of blog posts that best represent your expertise, not every post you have ever published. A model given fifteen tightly relevant links reads you as authoritative. A model given four hundred links reads you as noise. If a page would not make your top ten in a pitch to a new client, leave it out.

llms.txt is a highlight reel, not a sitemap. You already have a sitemap.xml for completeness. This file exists specifically because completeness is the wrong goal here.

Step 3: Write the descriptions

Each link gets a short, factual one-line description of what the page covers, not marketing copy. "Pricing and what is included in each tier" beats "See our incredible value." The model uses this line to decide relevance before it even opens the page, so treat it like a search snippet, not an ad.

Step 4: Host it correctly

Save the file as plain text at your site root so it resolves at /llms.txt with no redirect and a 200 status. Most static site setups, this one included, just need the file dropped in the public directory so the build serves it automatically. Confirm it loads directly in a browser before moving on.

Step 5: Verify it is working

There is no official verification dashboard, so check it the practical way: confirm the URL returns 200 and valid markdown, check your server logs after a week for requests to /llms.txt from known AI crawler user agents, and periodically test your buyer prompts in ChatGPT and Perplexity to see whether citations to your curated pages hold or improve. This is the same prompt-testing habit from tracking AI Overview visibility, pointed at one specific file.

Common mistakes to avoid

Listing every page on your site instead of curating. Letting the description drift out of sync with your actual brand description elsewhere. Forgetting to update it when you publish something genuinely important. And treating it as a one-time task rather than something you revisit each time your core content changes, the same way you would update a sitemap.

Keep it current

Add a reminder to review llms.txt whenever you publish a new pillar post or restructure your services, roughly quarterly for most sites. Pair it with clean schema markup so the curated pages you are pointing to are also individually well-structured, the two reinforce each other rather than one substituting for the other.

Who sets this up

The technical part takes under an hour. The part that actually matters, deciding which dozen pages represent you best and keeping that list current as your content evolves, is an editorial judgement call. It is the kind of small-but-compounding task a fractional marketing director folds into ongoing AI-search work rather than something that gets done once and forgotten.

FREQUENTLY ASKED

How do I implement llms.txt on my site?

Create a markdown file with an H1 brand name, a one-line description blockquote, and H2 sections grouping your most important pages with short factual descriptions. Save it as plain text at your site root so it resolves at /llms.txt with a 200 status, then verify it loads directly in a browser.

What pages should I include in llms.txt?

Your homepage, core service pages, pricing, and the handful of blog posts that best represent your expertise, not every page on your site. Curate to roughly ten to twenty links. A tightly relevant list reads as authoritative, a dump of every URL reads as noise.

How do I know if llms.txt is working?

Confirm the file returns a 200 status with valid markdown, check server logs for requests from known AI crawler user agents, and periodically run your buyer prompts through ChatGPT and Perplexity to see whether citations to your curated pages hold or improve over time.

How often should I update llms.txt?

Review it whenever you publish a genuinely important new page or restructure your services, roughly quarterly for most sites. Treat it like your sitemap: it needs periodic upkeep, not a one-time setup, or it drifts out of sync with what actually represents you.

RELATED SERVICES

Delivered in 22+ markets worldwide.

Want this done for your business?

Free audit. No pitch. 24-hour turnaround.

Get free audit →

PROOF THIS WORKS

MORE ARTICLES