LLMs.txt Generator
Build a spec-compliant llms.txt (and optional llms-full.txt) in your browser. Paste a base URL, import a sitemap, or add curated links. Output follows the llmstxt.org Markdown format that 2025–2026 AI crawlers and coding agents look for.
Nothing is uploaded to a server. fetch() for a sitemap runs from your browser; if CORS blocks it, paste the XML and we parse <loc> locally.
Generator
All fields stay on this device. The H1 site name is the only required llms.txt section.
sitemap.xml, so fetch() is blocked even though the file is public. Open the sitemap in a new tab, copy the XML, paste it below, then parse. This is a browser security limit, not a site error.
Manual link list
Each row becomes - [Title](URL): description under its H2 section. Sections map to llmstxt.org file lists (Docs, API, Blog, Legal, plus Optional for secondary links).
Live preview
How it works
- Identify the site. The generator writes an H1 from the site name (required by the spec), then an optional
> blockquotesummary and a details paragraph with no extra headings. - Curate links. Add Title / URL / one-line description rows and pick a section (Docs, API, Blog, Legal, or Optional). Use Base URL to resolve relative paths.
- Optional sitemap import. “Fetch sitemap” calls
fetch(sitemapUrl)from your browser. On a CORS failure you get an in-page note and a paste box. Pasted XML is parsed for every<loc>in a urlset or sitemapindex. Titles and sections are inferred from the path; you can edit or delete rows. - Two outputs.
llms.txtstays concise (curated rows plus a capped set of sitemap URLs). Tick the checkbox forllms-full.txtwith longer notes and a higher sitemap cap. Both stay Markdown lists:- [Name](url): notes. - Copy or download. Host the file yourself. This page never publishes it.
Practical caps (client-side, not a spec limit): llms.txt includes all curated rows plus up to 40 sitemap-sourced URLs; llms-full.txt includes all curated rows plus up to 200 sitemap-sourced URLs. Trim thin URLs before you go live — a sitemap is not a substitute for a curated map.
FAQ
What is an llms.txt file?
llms.txt is a Markdown file proposed at llmstxt.org. It sits at /llms.txt (or under a path such as /docs/llms.txt) and gives AI agents a short, curated map of a site: a title, an optional summary, optional details, and H2 sections of named links. It is meant to be small enough to fit in a context window.
How is llms.txt different from robots.txt?
robots.txt tells crawlers what they may fetch. llms.txt does not grant or deny access. It is complementary: when an agent is already allowed to read the site, llms.txt points it at the most useful pages. Use both — robots.txt for permission, llms.txt for guidance.
What is llms-full.txt?
llms-full.txt is a community convention, not a required llmstxt.org section. Sites publish a longer companion file with more links and longer notes. Host it at /llms-full.txt if you want a detailed index while keeping /llms.txt concise.
Where should I host llms.txt?
The common location is the site root: https://your-domain/llms.txt. The v2 proposal also allows a file at any path, covering URLs under that path (for example /docs/llms.txt). Serve it as text/plain or text/markdown. Optionally add rel="describedby" to pages it covers.
Do GPTBot and ClaudeBot read llms.txt?
GPTBot (OpenAI) and ClaudeBot (Anthropic) honor robots.txt first. llms.txt is a separate, on-demand map — useful when a coding or chat agent needs a guided path into your docs. llmstxt.org notes that OpenAI, Anthropic, and Gemini publish llms.txt on their own developer docs. This tool does not claim specific crawl rates or ranking effects.
Does this generator upload my site or sitemap?
No. Generation runs entirely in your browser. The optional Fetch sitemap button calls fetch() from your browser to the sitemap URL you typed. If that host blocks cross-origin reads (CORS), the request fails in-page and you can paste the XML instead. Nothing is sent to UtilForge.
What format does the llmstxt.org spec require?
Order matters: an H1 site name (the only required block), an optional blockquote summary, zero or more non-heading detail blocks, then zero or more H2 sections whose lists are - [Name](url): optional notes. An H2 titled Optional is the conventional place for secondary links. This generator follows that order.
Can I generate llms.txt from a sitemap?
Yes. Enter a sitemap URL and try Fetch sitemap, or paste sitemap XML. The tool reads <loc> values from urlset and sitemapindex documents, infers a title and section from the path, and adds editable rows. A sitemap is a full index; llms.txt should stay curated — delete thin URLs before you publish the file.