JPKCom DB

llms.txt Generator

What the llms.txt Generator does: build an llms.txt to spec v2, check it live including content type and links, with server snippets and a reference.

A file that explains your website to AI systems

An llms.txt is a small Markdown file in your domain's root — reachable at https://yourdomain.com/llms.txt — or under a subpath such as /docs/llms.txt, where it describes only that part of the site. It gives AI agents a curated, accurate entry point into your content: instead of blindly grazing a whole site, an agent reads the llms.txt like a table of contents and uses the link notes to decide which pages are worth fetching at all. Agents read it on demand, while they're helping someone — it is not a crawler or ranking signal. The llms.txt Generator builds exactly that file for you — and checks an existing one on request.

The problem it solves: llms.txt is Markdown with a fixed but easy-to-miss structure — an H1 title (the only required element), a blockquote summary, optional details, and thematic H2 sections holding link lists with notes. The tool follows version v2 (August 2026) of the specification. Written by hand, it's error-prone: a line of text before the H1, an empty URL, the wrong separator between link and note, or a parenthesis inside a URL, and agents and parsers read the file differently from what you intended. The generator takes the syntax off your hands: you fill in a form, the tool writes correct, properly escaped Markdown as you type and shows right below it what the validator makes of it.

The tool is for everyone who wants AI systems to capture and cite their site correctly: developers giving their docs or project site a clean llms.txt; content and SEO people who want to stay visible in generative search (GEO); and anyone who needs to quickly check their own or a third-party llms.txt. Everything runs in the browser — no account, no installation.

What the tool does

The llms.txt Generator has five tabs (the interface is in English):

  • Generator — the core: a form for the name, summary, details, sections with links and notes, H3 subsections, and an optional pointer to an llms-full.txt. The tool builds the finished llms.txt Markdown live in an editor on the right, with its check results right underneath. If you like, every link can point to the Markdown version of your page (page.html.md or page.md). Copy puts it on the clipboard, Download saves it as an llms.txt file; Open File or drag and drop loads an existing file back into the form — whatever the form can't represent as a field is kept as text. The Serving & discovery card produces matching configuration snippets for nginx, Apache, and HTML.
  • Check — the live checker/validator: enter a domain (example.com — /llms.txt is appended automatically), a directory (example.com/docs/), or a full URL, and the tool fetches the file and checks it against the llmstxt.org specification — along with the HTTP response itself: content type, character set, and HTML error pages served in place of a text file. You get errors, warnings, and info notes separately, a summary (title, summary, sections, links with notes, file size, reference parser), and an on-demand link check of the linked pages.
  • Examples — four real-world templates (open-source library, SaaS product, blog, docs site) you load into the generator with one click and adapt.
  • Tips — best practices: what makes a good llms.txt, what to avoid, how agents use it, where the file goes, and how big it may be.
  • Reference — the v2 format reference: a complete example, a field table (which element uses which syntax and whether it's required, recommended, optional, or not part of the spec at all), plus explanations of llms-full.txt, Markdown versions, and rel="describedby".

All in the browser — the generator needs no server at all

The generator works entirely client-side: as you type, JavaScript assembles the Markdown, the validator checks it, the editor shows it live, and your work in progress is saved locally in the browser (LocalStorage) — so it's still there on your next visit. A file you open is read by the browser itself, too. Nothing leaves your machine.

Only the Check tab needs the network, because for security reasons your browser can't fetch a third-party llms.txt directly (CORS). So a server-side JPKCom proxy retrieves the file (and, for the link check, the status and content type of each linked page) and hands it back to your browser; the actual validation again runs locally on your machine. The checked domain therefore sees a request from the JPKCom server, not your IP; internal and private addresses are blocked server-side. The two PHP endpoints behind this are not a public API — they're hardened exclusively for the tool.

Try it now

→ Open the llms.txt Generator — fill in the form, watch the Markdown and its check results build live, download the file, and place it in your web root. Or check an existing llms.txt in the Check tab. No account, free, right in the browser. If you'd like to see what a good file looks like first, the four templates in the Examples tab load ready-made structures into the generator.

The llms.txt is one building block of AI/GEO optimization — these tools complement it:

  • SEO & GEO Analyzer — inspects a whole page and rates the llms.txt as part of the GEO score (the llms.txt Present and llms.txt Valid Structure checks). Here you build the file; there you see how it feeds into the page's AI readiness — both tools share the same validator module.
  • Meta Tags Generator — clean titles and descriptions plus Open Graph tags — the latter also count towards the GEO score alongside the llms.txt (the "Open Graph Tags (≥ 4)" check).
  • robots.txt & Sitemap — build crawling rules and sitemap entries, including allow-listing AI crawlers.

There's more on the subpages: the manual with every field, every tab, and all validator rules, hands-on examples, and a collection of tips & tricks.