Make your site AI-readable, automatically.
llms.txt is a protocol standard for AI crawler access — a file at your domain root that helps AI engines discover and extract your content cleanly. SearchChamp's generator crawls your site, writes section-by-section summaries that match your Brand Voice, and regenerates automatically when you publish. It aids discoverability and extractability — not a citation lever or a ranking signal.
Credit card required · Cancel anytime · 7-day full access
Host the file at atlio.ai/llms.txt. AI search agents look for it at this exact path on your domain.
| Version | Generated | Status | Trigger |
|---|---|---|---|
| v3 | 2h ago | Published | Auto |
| v2 | 1 week ago | Superseded | Auto |
| v1 | 3 weeks ago | Superseded | Manual |
A map of your site, written for a language model
llms.txt is a plain-text file at your domain root that describes what your site contains — a curated, summarised map in prose an engine can lift from directly. SearchChamp also generates the extended /llms-full.txt, which carries the same map plus the full text of your articles and products appended underneath.
- Lives at yourdomain.com/llms.txt — the exact path AI agents look for
- Sections your site actually has: About, Documentation, Articles, Products
- A spec-compliant /llms-full.txt ships alongside it with the full text appended
We measured whether it works.
It didn't move the needle.
Most tools sell llms.txt as an AI-visibility lever. We had the data to check, so we checked — and published what we found rather than the pitch.
Across monitored accounts, a change in answer-engine citation rate within 90 days of publishing llms.txt was not distinguishable from noise.
SearchChamp AI Visibility panel — before/after cohort, 2025Most published llms.txt files are never fetched by AI crawlers, and no confirmed citation benefit was found.
Ahrefs — analysis of 137,000 sites, 2026So here is what we did about it
We'd rather tell you than sell you
Publish it anyway — because it costs you nothing
It is a proposed convention, not a ratified standard like robots.txt or sitemap.xml, and no engine is obliged to read it. What makes it worth publishing regardless is the asymmetry: it takes minutes to generate, nothing to keep current, and the engines that do use it get a clean, structured view of your site instead of guessing from your HTML. Small upside, near-zero cost — so the sensible move is to have one and stop thinking about it.
- Minutes to generate, and it maintains itself after that
- Free and unmetered — regenerating on publish costs no credits, ever
- The engines that do read it get a clean map instead of guessing
A proposed convention, not a ratified standard
Free generators hand you a skeleton. The summaries are the work.
Most llms.txt tools produce the structure and leave you to write every section summary yourself — which is the part that decides whether an engine can use the file at all. SearchChamp crawls your site, groups pages into the sections you actually have, and writes a concise summary for each one in your active Brand Voice, using the same trained profile as the content editor. Edit any section by hand and your override survives every future regeneration.
- Section-by-section summaries written for you, from a real crawl of your site
- Written in your trained Brand Voice — the same profile the editor uses, not a generic AI tone
- Edit a section and SearchChamp preserves your override on every refresh
The structure is the easy half
It regenerates when you publish — and never overwrites your edits
Publish new content and SearchChamp re-runs the section summaries within about 15 minutes, debounced so a burst of publishing collapses into one regeneration. Before every regeneration it fetches the file that is actually live and diffs it against the last version it pushed. If you have hand-edited the live file, it stops, flags it, and shows you the diff — the new version waits for your approval instead of overwriting your work.
- Regenerates within ~15 minutes of a publish, debounced to one run per site
- Fetches the live file first and diffs it before touching anything
- Hand-edited the live file? It holds the new version for review instead of overwriting
Your live file was edited by hand. Review the diff before this version replaces it.
It asks before it overwrites you
Every version kept, every change diffable
Each regeneration writes a new immutable version rather than replacing the last one, with a plain-English summary of what changed. Compare any two versions line by line, and roll back to an earlier one if a regeneration picked up something you didn't want.
- Immutable, append-only versions — nothing is overwritten in place
- A plain-English delta on every version, and a line-by-line diff between any two
- Roll back to an earlier version at any time
12 articles added, 3 retired products removed.
Nothing is overwritten in place
Four ways to get it live — and we're honest about which are automatic
WordPress publishes automatically through the same secure lane the rest of the platform uses. If you're headless or on a stack we can't write to, SearchChamp hosts the canonical file and gives you the redirect snippet to point your own root at it. There's a manual path with the exact file and placement steps, and a guided Webflow path — Webflow has no publish API for this, so that one hands you the file and the Designer steps rather than pretending it's automatic.
- WordPress — published automatically to your site root
- SearchChamp-hosted — a canonical URL plus ready-made redirect snippets
- Manual and Webflow — the exact file and placement steps, honestly labelled as manual
Published to your root through the secure publish lane.
TO YOUR SITE ROOTCanonical URL + redirect snippets (Cloudflare and others).
REDIRECT FROM YOUR ROOTFile content, placement steps, and a "mark as done" check.
MARK AS DONEDesigner steps, custom domains only — no publish API exists.
CUSTOM DOMAINS ONLYWE LABEL GUIDED PATHS AS GUIDED. THEY ARE NOT ONE-CLICK AND WE DON'T SAY THEY ARE.
Four paths · Two automatic · Two honest about being manual
"Published" isn't the same as "live"
After a publish, SearchChamp fetches your domain root and reads back what's actually there before it calls anything done. If the check fails it tells you precisely why — nothing at that URL, a timeout, a server error, an HTML page where a text file should be — rather than showing a green tick and hoping.
- The status only reaches "live & verified" after we fetch the file back from your domain
- Failure modes are named, not collapsed into one generic error
- The first lines of the live file are shown, so you can see what an engine would see
✓ Found a live llms.txt at this URL.
We read it back before we call it done
And then you can check whether a crawler ever fetched it
Since most llms.txt files are never fetched, the honest follow-up question is whether yours was. SearchChamp's AI Crawler Analytics records every AI-crawler hit on your site — which crawler, which path it requested, and whether the identity checks out against the vendor's published IP ranges — so the answer is something you can look up rather than assume.
- Every AI-crawler hit recorded with the path it actually requested
- Identity verified against each vendor's published IP ranges — spoofed agents surfaced as spoofed
- So "did anything read my llms.txt" becomes a lookup, not a guess
An answer, not an assumption
robots.txt, sitemap.xml, llms.txt — you want all three
They answer different questions and none of them replaces another. robots.txt says what a crawler may fetch. sitemap.xml says where your pages are. llms.txt says what they mean. Publishing llms.txt does not change what your sitemap or robots file are for, and it does not make either optional.
- robots.txt — permission. sitemap.xml — location. llms.txt — meaning
- llms.txt does not replace your sitemap, and it is not a ranking input
- Only robots.txt and sitemap.xml are ratified standards; llms.txt is a proposed convention
Three files · Three different questions
Generate one right now, without an account
The free generator crawls your domain and builds a spec-compliant file with 5 to 15 sections, emailed to you. No signup, no card, up to 5 runs per email address per day. It's the same crawl-and-summarise engine — the paid product adds the auto-refresh, the delivery paths, the versioning and the verification around it.
- No signup and no card — enter a domain, get the file by email
- 5 to 15 sections, spec-compliant, ready to deploy at your root
- Up to 5 runs per email address per day
No account · No card · Same engine
One URL. SearchChamp crawls your site and groups the pages into the sections you actually have
A concise summary per section, in your trained Brand Voice — your manual edits are preserved
Automatic to WordPress or SearchChamp-hosted; guided steps for manual and Webflow
Regenerates within ~15 minutes of a publish, diffed and verified, and never overwrites your edits
llms.txt, answered.
Make your site AI-readable, automatically
7-day free trial · credit card required · cancel anytime