Skip to content
50% OFF $299 $599
Lock in
§ 1.6.4 ARTICLE

llms.txt · validation

Validate llms.txt on Squarespace

Two minutes of validation catches every failure mode this cluster sees. The curl chain confirms the redirect and the body. Three free spec validators2 check the markdown against Jeremy Howard's specification1 — H1 present, blockquote optional, link integrity intact. Whether AI engines actually retrieve the file at answer time is harder to confirm, and the honest 2026 answer is covered below.

The curl chain that catches every install bug

Two curl invocations. The first fetches headers and follows redirects so you can confirm the chain — one 301 to /llms (or /s/llms.txt), then a 200 OK. The second fetches the body so you can confirm the manifest text returns and that the Markdown block on the Squarespace destination is rendering, not the Squarespace HTML chrome. If both pass, the install is mechanically correct.

bash The validation one-liner — copy, paste, replace the host 
 # 1) Inspect the redirect chain
curl -IL https://yoursite.com/llms.txt
# Expected: one HTTP/2 301, Location: /llms, then HTTP/2 200 OK # Failure modes: 404 (destination missing), 500 (page disabled), double 301 (page itself redirects) # 2) Fetch the body
curl -L https://yoursite.com/llms.txt
# Expected: markdown body with H1, optional blockquote, H2 sections # Failure mode: full Squarespace HTML page (Markdown block misconfigured)

On Windows without curl available, the same checks run with Invoke-WebRequest -Uri https://yoursite.com/llms.txt -MaximumRedirection 5. The redirect chain is reported in the History property; the body is in .Content.

Three spec validators worth running

Three free validators check the manifest against the official spec. Each checks for the required H1 — the one element the spec actually requires — plus the optional blockquote and the structural integrity of the H2 file-list links. Use at least two. One validator failing is usually a flag worth investigating; two validators failing on the same point is almost always a real bug in the manifest body.

  • llmstxtvalidator.dev2 — paste the URL or the file content. No registration. Checks syntax, structure, and spec compliance.
  • llmstxtchecker.net3 — URL-fetch validator. Checks for the required H1, optional blockquote, and link integrity in the H2 file lists.
  • Rankray LLMs.txt Checker4 — overlaps with the previous two and adds a missing-section detector.

All three run against a live URL — paste https://yoursite.com/llms.txt in. Because Squarespace serves the destination as text/html, some validators show a Content-Type warning even when the markdown body is correct. That warning is informational, not a failure. The hub5 covers the served-Content-Type caveat in detail.

Live signals from AI engines (what is checkable, what is not)

A clean curl chain plus a passing validator confirm the file is installed correctly. They do not confirm an AI engine is reading it. As of Q1 2026, no major engine — OpenAI, Anthropic, Google, Perplexity — has publicly committed to using third-party llms.txt as a retrieval or ranking signal at answer time. The realistic measurement check: confirm GPTBot, ClaudeBot, and PerplexityBot are reaching the URL in your server logs over a multi-week window, and treat any retrieval beyond that as bonus.

Squarespace's analytics panel does not surface user-agent strings at the file-request level, so the server-log check is partial on Squarespace alone. A practical proxy: enable a Cloudflare front, configure log forwarding, and watch for GPTBot, ClaudeBot, Claude-User, PerplexityBot, and Perplexity-User requests to /llms.txt over four to six weeks. If you see them, the file is at minimum being fetched. If you do not, the file still ships its standards-conformance benefit at zero ongoing cost.

The four failure modes a Squarespace install actually hits

The same four failures account for almost every llms.txt install bug on Squarespace. The fix for each is one click or one URL Mapping edit. Run the curl chain and the validators first; the failure mode is usually visible inside thirty seconds.

  • 404 on /llms.txt after the URL Mapping is saved. The destination page does not exist or is set to Disabled. Open the /llms page settings and confirm it is enabled and published.
  • Body returns wrapped in full Squarespace HTML chrome. The destination is a regular page with header and footer rather than a Markdown block on a blank layout. Edit the page, remove all other blocks, leave only the Markdown block.
  • Double 301 in the curl chain. The /llms page itself has a redirect set in page settings. Open page settings and clear the URL forwarding field.
  • Validator flags missing H1. The Markdown block opens with ## or plain text. The spec requires exactly one #-prefix H1 as the first line. Edit the body to start with # Site Name.

Once curl and one validator both pass, the install is done. Move on to the AI Crawlers cluster if you have not already — llms.txt is one signal; crawler-access settings are the other half of the same picture.

Keep reading · § 1.6