llms.txt Validator

What does the llms.txt validator check?

The validator fetches your live site and runs the same llms.txt checks as a full AgentGrade scan:

• llms.txt found — the file exists at /llms.txt, starts with an H1 naming the service on line one (the only element the spec strictly requires), and follows with the blockquote summary the spec recommends.
• Markdown structure — the file contains something an agent can actually parse: a list of markdown links to real pages, a fenced code block with a sample request, or named operational sections like ## Endpoints or ## Authentication.
• llms-full.txt found — the companion corpus file exists, so non-browsing LLMs can ingest your content in one fetch.
• Discovery links — your HTML declares <link rel="alternate" type="text/plain" href="/llms.txt"> so crawlers connect the file to your pages.

What is llms.txt?

/llms.txt is a markdown file at your domain root that tells AI agents what your site offers and how to use it — an operating manual, not a sitemap. When an agent lands on a site with a good llms.txt, it skips the scraping phase entirely: one fetch tells it what the service does, which endpoints exist, and how to authenticate. Docs platforms like Mintlify now generate it automatically for every site they host. The format is specified at llmstxt.org; see the AgentGrade llms.txt guide for a deeper walkthrough.

A valid llms.txt example

# example.com

> One sentence describing what the service does and when an agent should use it.

## Endpoints

- `POST /api/v1/things` — Create a thing. Body: `{ "name": "widget", "qty": 5 }`.
- `GET /api/v1/things/{id}` — Fetch a thing by id.

## Authentication

Bearer tokens issued at `/api/keys`. Send as `Authorization: Bearer <token>`.

## Examples

    curl -X POST https://example.com/api/v1/things \
      -H "Authorization: Bearer $TOKEN" \
      -d '{"name":"widget","qty":5}'

Common llms.txt mistakes

The sitemap dump

A list of bare URLs with no titles or descriptions. Parsers can't extract meaning from naked links — each bullet should be a markdown link with a short description: - [API Reference](/docs/api) — every endpoint with parameters.

robots.txt syntax

Files written with # comment lines and directive-style entries. llms.txt is markdown, not robots.txt — # starts a heading, and the first one must be your service name.

Plain-text bullets

- About and - Twitter tell an agent nothing and link nowhere. Every list item needs a real destination and a reason to follow it.

Invented SPA paths

Listing client-side router paths that all return the same HTML shell. Only list URLs that return real content when fetched directly — agents verify, and dead paths waste their budget and your credibility.

No example requests

A fenced code block with one working curl makes your file executable documentation instead of prose. It's also one of the three structures that pass the markdown check on its own.

Frequently asked questions

Is llms.txt required for AI agents to use my site?

No — agents can still crawl HTML without it. But it's the cheapest agent-readiness win available: one static file, about ten minutes of work, that saves every visiting agent from scraping your site to figure out what it does.

What's the difference between llms.txt and llms-full.txt?

llms.txt is the index — a short manual pointing at your key URLs. llms-full.txt is the corpus — your site's full text concatenated into one file for RAG pipelines and non-browsing models.

Does llms.txt affect Google rankings?

It isn't a conventional ranking factor. Its value is agent-side: LLM-driven tools and assistants that fetch it get an accurate picture of your service, which is what determines whether they can use — and recommend — your site.

How is this different from a full AgentGrade scan?

This page runs the llms.txt slice of the full scan. The complete scan covers 70+ checks across payments, MCP, OpenAPI, content negotiation, and infrastructure — with the same instant, no-signup report.