How do I validate an llms.txt file?

Paste the file into a validator such as the one above. It checks the title, summary, sections, link format, URL scheme, duplicates and size, and reports problems with line numbers. It cannot test whether links load, so open them or run a crawler too.

What are the most common syntax errors in llms.txt?

The most common errors are a missing or duplicate # title, list items not written as - [Name](url): note, relative or http:// URLs, empty sections, and more than one Optional section.

What is the required format for links in llms.txt?

Write each link as - [Title](https://example.com/page): Optional note. The link is required, and the note after the colon is optional but recommended. Use absolute https URLs.

Can an llms.txt file have multiple H1 titles?

No. Use exactly one # title at the top and ## for sections. A second H1 can confuse parsers, and this validator reports it as an error.

What is the difference between llms.txt and llms-full.txt?

llms.txt is an index of links with short notes. llms-full.txt is a common companion convention, not part of the core proposal, that holds the full text of key pages in one file.

What MIME type should be used for serving llms.txt?

The proposal does not mandate one. Serve it as text/plain; charset=utf-8 or text/markdown; charset=utf-8 so browsers show the text instead of downloading it or rendering it as HTML.

Where should the llms.txt file be hosted?

At the root of your domain, for example https://example.com/llms.txt. The proposal also allows a file in a subpath for a docs section, and each subdomain needs its own file.

How do AI agents like Cursor or Windsurf parse llms.txt?

I could not confirm that Cursor or Windsurf read llms.txt automatically. Both can take documentation from a URL or pasted text as context, so give them your llms.txt link or its contents and test the result.

LLMs.txt Validator: Check Markdown Syntax

llms.txt validator

Save on Google

Free online llms.txt checker. Paste your file and see errors, warnings and tips with line numbers. Everything runs in your browser. Nothing is uploaded.

Paste llms.txt

Results & Output

The output box shows your file cleaned up: LF line breaks, no trailing spaces, one blank line between blocks and - bullets.

What does an llms.txt validator check?

The validator checks that your llms.txt follows the proposed Markdown format: one # title, an optional > summary, and ## sections of links written as - [Name](url): note. It flags missing or broken pieces so you can fix them before uploading the file to /llms.txt.

llms.txt validation rules: 70 checks and specifications

Severity is a recommended level. The validator above checks the title, summary, sections, links, URLs and size. Server and agent rows are reference only. Rules of thumb are labelled as such, and rows marked Unverified are items I could not confirm.

Showing 70 of 70 validation rules & specifications

Validation Rule / Parser Query	Specification Standard & Error Fix	Severity Level	Scope / Category
Missing single H1 title	Exactly one `# Title` line must come first. Fix: add `# Your site name` at the top.	Fatal Error	Anatomy & H1
Multiple H1 headings detected	Only one `# Title` is allowed. Fix: change later titles to `## Section`.	Fatal Error	Anatomy & H1
Non-UTF-8 encoding detected	Save as UTF-8. Curly quotes and accented letters are valid UTF-8, but bytes that are not valid UTF-8 can garble the file or make a tool reject it. Fix: re-save as UTF-8.	Fatal Error	Server & Encoding
Trailing spaces in H1 header	Trailing spaces after `# Title` can leak into the parsed title in some tools. Fix: trim the line.	Warning	Anatomy & H1
Missing blockquote summary	The proposal recommends `> Short project summary` right after the H1. Fix: add one or two factual sentences.	Warning	Summary Block
Multi-line blockquote syntax	Start every line of a multi-line summary with `>`, or keep the summary on one line.	Best Practice	Summary Block
Summary preceding H1	The `# Title` must come before any `> summary`. Fix: move the title to line 1.	Fatal Error	Summary Block
Empty file validation	A file with 0 bytes or only whitespace has no title or links. Fix: add a title and at least one link.	Fatal Error	Anatomy & H1
Case sensitivity of filename	Use lowercase `llms.txt`. On a case-sensitive server `LLMS.txt` returns 404, so tools never find it.	Fatal Error	Server & Encoding
Default root path validation	Host it at the root of the domain: `https://example.com/llms.txt`.	Spec Requirement	Server & Encoding
Subdirectory placement validation	The proposal also allows a file in a subpath for an isolated docs section, for example `https://example.com/docs/llms.txt`.	Best Practice	Server & Encoding
HTML tags in Markdown	Use plain Markdown. HTML tags are not part of the format and may show as raw text. This validator flags them.	Warning	Anatomy & H1
Leading blank lines before H1	This validator ignores blank lines before the title, but a strict parser may expect the H1 on line 1. Fix: remove them.	Warning	Anatomy & H1
BOM (byte order mark) presence	A UTF-8 BOM can break strict Markdown parsers or show as stray characters before the title. Fix: save as UTF-8 without BOM.	Warning	Server & Encoding
Line ending consistency (LF vs CRLF)	Prefer Unix LF line breaks. This validator accepts CRLF too.	Best Practice	Server & Encoding
Unrecognized H2 section header	Section names are free-form. Clear names such as `## Docs`, `## API Reference` or `## Guides` help. Only `## Optional` has a defined meaning.	Best Practice	Sections
H3 or H4 sub-headings usage	Headings below `##` are not part of the format. This validator reports them as tips. Fix: use flat `##` sections.	Best Practice	Sections
Missing section headers	Links need a `##` section. Fix: add `## Docs` above the first list. This validator warns about list items before the first section.	Warning	Sections
Duplicate section names	Repeating a `## Section` name makes the file confusing. Fix: merge or rename the sections.	Warning	Sections
The `## Optional` section rule	Use one `## Optional` section, last in the file, for links a tool may skip when it needs a shorter context.	Spec Requirement	Sections
Empty section headers	A `##` heading with no links is flagged by this validator. Fix: add links or delete the heading.	Warning	Sections
Narrative paragraph placement	Plain paragraphs are allowed between the summary and the first `##` section. Text inside a section is reported as a tip.	Spec Requirement	Sections
Excessive paragraph length	Keep paragraphs short, as a rule of thumb under about 500 characters, because long text uses token budget.	Best Practice	Sections
Code fences in root file	Avoid fenced code blocks (three backticks) in the index. Put long examples in `/llms-full.txt`.	Best Practice	Sections
Horizontal rule usage (---)	A `---` rule is not part of the format and adds nothing. Fix: remove it.	Best Practice	Sections
Bulleted list type mismatch	Use hyphens: `- [Name](url)`. This validator accepts `*` too, but the proposal examples use `-`.	Best Practice	Sections
Numbered lists for links	Use bullet lists, not `1.` lists, so each link line matches the standard pattern.	Best Practice	Sections
Nested lists inside links	Keep one flat level. Indented sub-bullets can be missed by simple link extractors.	Best Practice	Sections
Canonical link syntax	`- [Title](https://example.com/page): Optional notes`	Spec Requirement	Links & URLs
Missing link title bracket	`-[Title(url)` is broken. Fix: `- [Title](https://example.com/page)`.	Fatal Error	Links & URLs
Missing URL parentheses	`[Title]https://example.com` is broken. Fix: wrap the URL: `[Title](https://example.com)`.	Fatal Error	Links & URLs
Missing colon before note	Write `[Title](url): note`. Note text after the link without the colon makes the line invalid in this validator.	Fatal Error	Links & URLs
Relative vs absolute URLs	Use absolute URLs such as `https://example.com/docs` so a tool needs no base domain.	Best Practice	Links & URLs
Missing protocol in URL	`example.com/doc` and `/doc` are not absolute. Fix: start the URL with `https://`. This validator reports it as an error.	Fatal Error	Links & URLs
Insecure HTTP URLs	`http://` links are flagged. Fix: use `https://` where the site supports it.	Warning	Links & URLs
Dead or 404 links	This validator cannot check them, because a browser page cannot request other sites. Test links with a crawler and fix any `404`.	Warning	Links & URLs
Redirect chains (301/302)	Link to the final URL. Redirects cost extra requests, and a tool may stop following them.	Warning	Links & URLs
Trailing slash mismatches	`/docs` and `/docs/` can be different URLs. Use the exact form your server returns with `200`.	Warning	Links & URLs
URL fragment identifiers (#hash)	A `#section` part points inside one page. Many tools fetch the whole page, so the fragment may be ignored.	Best Practice	Links & URLs
Query parameters in links	Remove tracking parameters such as `?utm_source=` and keep only parameters that change the content.	Warning	Links & URLs
Linking HTML instead of raw Markdown	Where you have clean Markdown versions, link them, for example `page.md`. They use fewer tokens than HTML.	Best Practice	Links & URLs
Duplicate URL entries	List each URL once. This validator flags repeats and shows the line of the first use.	Warning	Links & URLs
File size exceeding 50 KB	There is no official limit. As a rule of thumb, keep `/llms.txt` under about 50 KB. This validator warns above 20,000 characters.	Best Practice	Size & Tokens
Token count exceeding 10,000 tokens	A practical target is under about 10,000 tokens, at roughly 4 characters per token. Move depth into `/llms-full.txt`.	Best Practice	Size & Tokens
Excessive link density	A very long list dilutes your key pages and tools may truncate it. As a rule of thumb, stay near a few dozen links and use `## Optional` for the rest.	Best Practice	Size & Tokens
Vague link descriptions	Notes such as “docs” or “click here” tell a model nothing. Fix: say what the page covers.	Best Practice	Size & Tokens
Optimal link description length	As a rule of thumb, keep notes between 5 and 25 words.	Best Practice	Size & Tokens
Difference between /llms.txt and /llms-full.txt	`/llms.txt` is an index of links. `/llms-full.txt` is a common companion convention that holds full page text in one file.	Best Practice	Size & Tokens
Circular linking to /llms.txt	The file does not need to link to itself. Remove such a link to save space.	Best Practice	Links & URLs
Binary file linking	Links to PDFs, ZIPs or images give a text-only tool nothing to read. Link a text or Markdown version instead.	Warning	Links & URLs
Authenticated or paywalled URLs	Links that return `401` or `403` cannot be read by public tools. Remove them or provide a public version.	Warning	Links & URLs
Rate-limited endpoints (429)	A docs host that returns `429` to automated requests can stop tools from reading your links. Check your rate limits.	Warning	Links & URLs
Caching headers (Cache-Control)	`Cache-Control: public, max-age=86400` caches the file for one day. Use less if you update it often.	Best Practice	Server & Encoding
Compression support (Gzip/Brotli)	Serve it with `Content-Encoding: gzip` or `br`. Text compresses well and clients decompress it automatically.	Best Practice	Server & Encoding
Correct MIME type header	The proposal does not mandate one. Use `text/plain; charset=utf-8` or `text/markdown; charset=utf-8`.	Best Practice	Server & Encoding
Incorrect MIME type (text/html)	A file served as `text/html` may render as a web page or be rejected. Fix: return a text type.	Warning	Server & Encoding
Missing CORS header	`Access-Control-Allow-Origin: *` is only needed when browser-based tools on other origins fetch the file. Server-side tools do not need it.	Best Practice	Server & Encoding
Robots.txt disallow on root	If `robots.txt` disallows `/llms.txt` or `/`, compliant crawlers will not fetch the file.	Warning	Server & Encoding
Robots.txt explicit allow rule	Add `Allow: /llms.txt` inside a group that otherwise blocks the site.	Best Practice	Server & Encoding
Nginx location block configuration	`location = /llms.txt { default_type text/markdown; charset utf-8; }`	Best Practice	Server & Encoding
Apache .htaccess configuration	`<Files "llms.txt"> ForceType "text/markdown; charset=utf-8" </Files>` Avoid `AddType ... .txt`, which changes every `.txt` file.	Best Practice	Server & Encoding
Cloudflare edge rule compatibility	Cloudflare minification and Rocket Loader only touch HTML, CSS and JS. Still check the response headers and make sure no WAF or bot rule blocks `/llms.txt`.	Best Practice	Server & Encoding
Next.js static route generation	Put the file in `public/llms.txt`, or return text from a Route Handler such as `app/llms.txt/route.ts`.	Best Practice	Maintenance
Astro documentation integration	Generate it from your docs collection with an endpoint such as `src/pages/llms.txt.ts`.	Best Practice	Maintenance
Cursor IDE automated discovery	I could not confirm that Cursor reads `/llms.txt` automatically. Cursor can add documentation from a URL, so test with your own link.	Unverified	Agent Ingestion
Windsurf context ingestion	I could not confirm a built-in `llms.txt` parser in Windsurf. Paste the file or its URL as context and check Windsurf’s docs.	Unverified	Agent Ingestion
LangChain document loader validation	I am not aware of a loader named `LLMSTxtLoader`. Load the URL with a generic web loader, then split on `##` headings.	Unverified	Agent Ingestion
LlamaIndex reader compliance	Download the file and read it with a Markdown reader, or load the URL with a web page reader. Check the current LlamaIndex docs.	Unverified	Agent Ingestion
Automated CI/CD syntax testing	Run a validation script in GitHub Actions before deploy, and fail the build on errors or on a listed link returning `404`.	Best Practice	Maintenance
Syncing llms.txt with Git releases	Regenerate the link index when you publish a new version, and move old versions to `## Optional`.	Best Practice	Maintenance

llms.txt validator FAQ

Question	Answer
How do I validate an llms.txt file?	Paste the file into a validator such as the one above. It checks the title, summary, sections, link format, URL scheme, duplicates and size, and reports problems with line numbers. It cannot test whether links load, so open them or run a crawler too.
What are the most common syntax errors in llms.txt?	The most common errors are a missing or duplicate # title, list items not written as - [Name](url): note, relative or http:// URLs, empty sections, and more than one Optional section.
What is the required format for links in llms.txt?	Write each link as - [Title](https://example.com/page): Optional note. The link is required, and the note after the colon is optional but recommended. Use absolute https URLs.
Can an llms.txt file have multiple H1 titles?	No. Use exactly one # title at the top and ## for sections. A second H1 can confuse parsers, and this validator reports it as an error.
What is the difference between llms.txt and llms-full.txt?	llms.txt is an index of links with short notes. llms-full.txt is a common companion convention, not part of the core proposal, that holds the full text of key pages in one file.
What MIME type should be used for serving llms.txt?	The proposal does not mandate one. Serve it as text/plain; charset=utf-8 or text/markdown; charset=utf-8 so browsers show the text instead of downloading it or rendering it as HTML.
Where should the llms.txt file be hosted?	At the root of your domain, for example https://example.com/llms.txt. The proposal also allows a file in a subpath for a docs section, and each subdomain needs its own file.
How do AI agents like Cursor or Windsurf parse llms.txt?	I could not confirm that Cursor or Windsurf read llms.txt automatically. Both can take documentation from a URL or pasted text as context, so give them your llms.txt link or its contents and test the result.