llms.txt Generator – Make Your Site Citable by ChatGPT, Perplexity & Claude
A plain-text file at the root of your domain that gives large language models a curated map of your best content, in a format they can quote.
Generate yours automaticallyWhat is llms.txt?
llms.txt is a Markdown file placed at the root of a domain (for example, semoptimiser.com/llms.txt) that gives large language models a curated, quotable summary of the site and links to the canonical Markdown versions of key pages.
The format was proposed by Jeremy Howard of Answer.AI in September 2024, modelled on robots.txt and sitemap.xml but aimed at inference-time retrieval rather than crawling.
A valid llms.txt starts with a single H1 (the site name), followed by a blockquote one-paragraph summary, then H2 sections listing your best URLs as Markdown links with short descriptions.
How LLMs use it
When an LLM agent needs current information about a domain – a shopping query, a product comparison, a support question – it can fetch /llms.txt and read the site's own summary of what it contains. The agent then follows the linked Markdown URLs (.md versions of each page) to retrieve clean, layout-free copy that fits in a context window.
Retrieval-augmented generation (RAG) pipelines that index the open web commonly prefer llms.txt when it is present because it removes ambiguity about what content on the domain is canonical.
As of mid-2026 no major model provider has publicly committed to reading llms.txt automatically at inference time. Adoption today is publisher-driven and tool-driven – agents pointed at the file explicitly, and frameworks that build indexes. The cost of publishing is near zero and it future-proofs the site as adoption grows.
Annotated example
A minimal, valid llms.txt for a SaaS product:
# SEMOptimiser # H1: the site name (required)
> An open, API-first SEO/SEM intelligence # Blockquote: one-paragraph
> platform with native Model Context Protocol # summary of the site.
> (MCP) support. Keyword research, rank tracking, # LLMs will quote this
> competitor analysis, site audit and reporting. # verbatim as a card.
## Docs # H2: named section
- [Getting started](https://semoptimiser.com/docs/getting-started.md): 5-minute setup for a new workspace.
- [Rank tracker guide](https://semoptimiser.com/docs/rank-tracker.md): daily SERP checks with historical charts.
- [API reference](https://semoptimiser.com/docs/api.md): REST + MCP endpoints.
## Product
- [Pricing](https://semoptimiser.com/pricing.md): Free, Marketing $49/mo, Management $99/mo, Enterprise custom.
- [MCP integration](https://semoptimiser.com/mcp.md): connect Claude, Cursor, and any MCP host.
## Optional # Optional: lower-priority
# links an LLM may skip
- [Changelog](https://semoptimiser.com/changelog.md): weekly release notes.
- [Brand assets](https://semoptimiser.com/brand.md): logos and colour tokens.
Spec – every field explained
- H1 site name
- Exactly one # heading, at the very top. This is your brand or site name. Required.
- Blockquote summary
- A single > blockquote paragraph after the H1. One to three sentences. Factual, no marketing adjectives. This is what LLMs quote in citation cards. Required.
- H2 sections
- Group links under ## headings. Common groups: Docs, Product, Blog, API, Company. Order matters – the first sections are read first when context budget is tight.
- Link items
- Each item is - [Title](URL): description. Prefer .md URLs. Description is one sentence and appears next to the URL in retrieval results.
- Optional section
- A final ## Optional heading holds links that are useful but skippable. LLMs with a tight budget can drop this section.
- File location
- Must be served at the domain root path /llms.txt. Subdomains have their own. Redirects are not honoured.
- Content type
- Serve as text/markdown; charset=utf-8 (or text/plain). 200 status. No authentication.
Checklist for a good llms.txt
- 1
Draft the H1 and blurb
The first line is # SiteName. The next block is a > quoted one-paragraph summary of what the site does. This is what an LLM will paste into a summary card. Write it like a Wikipedia lede: factual, no marketing language.
- 2
Group your best URLs
Use ## section headings (Docs, Product, Pricing, Blog, API). Under each heading, list links as - [Title](https://yoursite.com/path.md): one-sentence description. Use the .md version of each URL when available.
- 3
Add an Optional section
The ## Optional section at the end holds links that are useful but not essential. LLMs with a small context budget can skip these safely.
- 4
Publish an .md per page
For every URL listed in llms.txt, publish a clean Markdown copy at the same path with .md appended. Strip navigation and ads. Keep front-matter minimal.
- 5
Serve at the domain root
Deploy to /llms.txt with Content-Type: text/markdown; charset=utf-8 and a 200 status. No auth, no redirects, no client-side rendering.
- 6
Validate and monitor
Fetch the file with curl to confirm it renders as plain text. Add it to your monitoring so a broken deploy does not silently 404 it.
FAQ
- What is llms.txt?
- llms.txt is a plain-text file placed at the root of a domain (yoursite.com/llms.txt) that gives large language models a curated, quotable summary of the site and links to the canonical Markdown versions of key pages. It is a proposal by Jeremy Howard (Answer.AI) modelled on robots.txt and sitemap.xml, but aimed at inference-time retrieval rather than crawling.
- Which LLMs respect llms.txt?
- As of mid-2026 no major LLM provider has publicly committed to reading llms.txt at inference time. Adoption today is on the publisher side: agents that browse the web (Perplexity, ChatGPT Search, Claude with tools) can be pointed to it explicitly, and retrieval frameworks that build local indexes commonly prefer it when present. Publishing it costs nothing and future-proofs the site.
- Is llms.txt the same as robots.txt?
- No. robots.txt tells crawlers which URLs they may fetch. llms.txt tells LLMs what your site is about and where the clean, machine-readable versions of your best content live. The two files coexist.
- Where do I put llms.txt?
- At the domain root: https://yoursite.com/llms.txt. Serve it as text/plain or text/markdown with a 200 status. No authentication, no redirects.
- Should each page have an .md version too?
- The convention pairs llms.txt with a clean Markdown version of every page at page-url.md. So /pricing has /pricing.md alongside it. LLMs prefer the .md because it strips navigation, ads and layout noise.
- Does llms.txt help SEO?
- Not directly for classic Google ranking. It helps AI Overviews, Perplexity, ChatGPT Search and Claude cite you accurately, which is the SEO surface that is growing. Treat it as AEO (answer-engine optimisation), not SEO.
One platform. Five fewer subscriptions.
Audit, index and rank –
without a fifth subscription.
Join 1,200+ agencies and in-house teams using SEMOptimiser to replace Semrush, Ahrefs, GA4 add-ons and rank trackers – with one workflow that actually ships fixes.