What is llms.txt? How the file works and who actually reads it

Editorial team · updated 24 September 2026 · facts checked 24 September 2026

llms.txt is a Markdown file placed at /llms.txt that gives language models a short, curated map of a website: a title, a one-line summary and lists of links to pages written for machines. Jeremy Howard proposed it at llmstxt.org on 3 September 2024; version 2 of the proposal followed in August 2026.

It is a proposal, not an IETF or W3C standard. Coding agents read it when a user, a rule or a line on a docs page sends them there. Google, OpenAI, Anthropic and Perplexity don't document it as a crawler input.

How to write an llms.txt file

The proposal fixes the section order; only the first is required:

1. An H1 with the name of the site or project.
2. A blockquote with a short summary.
3. Optional paragraphs or lists, without headings.
4. H2 sections holding file lists: - [name](url), optionally followed by : notes.

An Optional section, by convention, holds links an agent can skip. The file can sit at the root or under a path such as /docs/llms.txt, covering the pages beneath it, and the most specific file wins. An illustrative file for a made-up API:

# Example SMS API

> REST API for SMS and one-time codes. Base URL https://api.example.com/v1, Bearer API keys, JSON bodies.

Server-side calls only; webhooks are signed with HMAC-SHA256.

## Docs

- [Send an SMS](https://example.com/docs/send-sms.md): POST /messages, parameters, error codes
- [Webhooks](https://example.com/docs/webhooks.md): delivery receipts and signature checks

## Optional

- [Changelog](https://example.com/docs/changelog.md)

Link Markdown pages, not HTML, and keep the file short. The proposal's test: ask an agent about your product, giving it only your llms.txt as a starting point. Mintlify, GitBook and the WordPress plugins Yoast and AIOSEO can generate one.

llms-full.txt and Markdown page versions

The proposal asks for a clean Markdown copy of pages agents might need, at the same URL plus .md (page.html.md, or page.md since v2). Version 2 adds discovery hints: rel="alternate" type="text/markdown" for a page's Markdown twin, rel="describedby" for the llms.txt covering it. Mintlify uses rel="llms-txt" instead, plus a /.well-known/ copy, a location v2 decided against, so discovery isn't settled.

llms-full.txt is not in the proposal but a docs-platform convention: Mintlify describes it as your entire documentation in one file. Those get big: on 24 September 2026 Bird's was 2.5 MB, Resend's 2.2 MB and Plivo's 4.3 MB. Telnyx splits its own by product area because the combined corpus is "far past single-request agent limits".

llms.txt vs robots.txt vs sitemap.xml

llms.txt allows or blocks nothing. To keep an AI crawler out, use robots.txt.

Do AI crawlers use llms.txt?

Not according to their operators, on 24 September 2026:

• Google's AI features guide, updated 10 December 2025, says "You don't need to create new machine readable files, AI text files, or markup" to appear in AI Overviews or AI Mode.
• OpenAI, Anthropic and Perplexity describe robots.txt as the control on their crawler pages (OpenAI, Anthropic, Perplexity); none says its crawlers read llms.txt.

Yet OpenAI and Anthropic publish one for their developer docs, and OpenAI's docs pages point agents to it. Chrome's Lighthouse audit, updated 5 May 2026, flags a server error when fetching the file, while a missing file is "Not Applicable" because providing one is "optional at the moment". Treat it as documentation for agents, not an SEO lever.

Which email, SMS and WhatsApp APIs publish llms.txt

In our AI readiness dataset, checked 23 and 24 September 2026, 24 of the 27 providers publish an llms.txt for their developer docs; Twilio and SendGrid share one. Blip and Zenvia have only SEO-plugin files for their marketing sites, with no API docs, and Botmaker's returned HTTP 500. Full-text files re-fetched 24 September 2026; alphabetical within each group:

• Index, full-text export and Markdown pages: 360dialog (paginated), Amazon SES and AWS End User Messaging (a single full.md), Bird, Mailtrap, Plivo, Resend, Telnyx (split by product) and Treble.
• Index and Markdown pages: Brevo (its llms-full.txt URL returns the index), ClickSend, Gupshup, Infobip, Mailgun, Mailjet, Postmark (Markdown on Accept: text/markdown), Sinch (its standalone SMS API docs are missing from the index), SMTP2GO, Twilio, Twilio SendGrid and WATI.
• Index only: SendPulse (links an OpenAPI file per product; its llms-full.txt is a 7.7 KB service list), Textmagic and Vonage.

Our transactional email, SMS, OTP and WhatsApp rankings give an llms.txt up to 2.5 of 100 Developer & AI points, an official remote MCP server up to 27.5 (methodology).

Point Claude Code or Cursor at a provider's llms.txt

Give the agent the URL and say what to read (illustrative, not run):

Read https://resend.com/docs/llms.txt and the .md pages it links on sending and webhooks, then write the send function.

• Claude Code fetches with its WebFetch tool, which truncates large pages and usually hands Claude a small model's answer, not the raw text. Even indexes can be large (Twilio's was 380 KB on 24 September 2026), so name the pages you want; for raw text the docs suggest curl through Bash. An allow rule like WebFetch(domain:resend.com) skips the approval prompt, and a line in CLAUDE.md makes the habit permanent.
• Cursor's agent can search the web, drive a browser and run shell commands, but on 24 September 2026 its @-mention list had no docs entry. Put the URL in a project rule or AGENTS.md.

Read the file before trusting it; Mintlify lets docs owners add an "Agent Instructions" block. For account access rather than docs, see MCP server and our email MCP comparison.

PingMyUsers' own llms.txt

This site publishes /llms.txt, linking its read-only MCP server and REST API, dataset, rankings, provider cards and guides. It serves no Markdown pages; agents query the API instead.

Related guides

• Email MCP servers compared
• AI readiness dataset: MCP, skills, llms.txt and OpenAPI for 27 providers
• This directory's API and MCP server

About this guide

The PingMyUsers editorial team wrote this entry for developers who wire messaging APIs into AI coding tools. Sensaria AG in Switzerland operates the directory; providers don't pay for placement, and none reviewed this page. Report errors, with a source, to contact@sensaria.ch.

Methodology

On 24 September 2026 we read the proposal and its change log, Mintlify's docs, RFC 9309, sitemaps.org, the crawler or AI-search pages of four companies, the Lighthouse audit and the Claude Code and Cursor docs, and fetched every provider's llms.txt and llms-full.txt. Groups start from the directory's dataset. We ran no agent against these files.

Last updated

24 September 2026: first version. Next re-check: March 2027, or sooner if the proposal or a crawler policy changes.

Frequently asked questions

Is llms.txt an official standard?

No. It is a proposal, at version 2 since August 2026, with no IETF or W3C status.

Does llms.txt help SEO?

Not in Google Search: AI Overviews and AI Mode need no AI text files, Google says. It helps agents that read your docs.

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

llms.txt is a short index of links; llms-full.txt, a docs-platform convention, holds the whole documentation in one file.

Where do I put llms.txt?

At the root, /llms.txt, or under a path such as /docs/llms.txt to cover only that section.

Do ChatGPT and Claude read llms.txt?

Their crawler pages don't say so, but given the URL they can fetch it like any page.