@lacspace/llms-txt

Generate & parse llms.txt and llms-full.txt — the llmstxt.org standard.

1.5.0 — link text is escaped by default, and multi-line summaries survive

Two bugs that both lost content from the file an LLM reads:

• A title containing [ or ] dropped the whole link. Escaping was opt-in, so the default output for an ordinary title was - [Guide [v2]](/g) — not a parseable Markdown link. parseLlmsTxt() read it as no link at all, and the page silently vanished. Escaping is now on by default; it only ever touches \ [ ] ( ), so titles without them are byte-for-byte unchanged. Pass { escape: false } for the old behaviour.
• A multi-line summary became two fields. > ${summary} quoted only the first line, so the rest fell out of the blockquote and parsed back as details. Every line is quoted now.

parseLlmsTxt() learned to read escaped link text (the old regex stopped at the first ]), and unescapeLlmsText() is exported as the exact inverse of escapeLlmsText(). llmsTxt() → parseLlmsTxt() now round-trips exactly for brackets, parens, backslashes, notes, the ## Optional block and multi-line summaries. Older hand-written unescaped files still parse.

llms.txt is a Markdown file at your site root that gives LLMs a curated map of your most useful pages; llms-full.txt inlines the full content so a model can read everything in one request. This builds (and parses) both — the SEO layer for the AI era.

• 📄 llmsTxt() — H1 title, blockquote summary, linked sections, ## Optional block
• 📚 llmsFullTxt() — full content inlined for one-shot ingestion
• 🗺️ llmsFromPages() — build BOTH files from one list of pages (title + url + content)
• ✅ validateLlmsTxt() — check the required H1 + summary shape, report issues
• 🔁 parseLlmsTxt() — read an existing file back into structured data
• ⚡ Zero dependencies · 🌍 isomorphic · 📦 ESM + CJS · fully typed

New in 1.4.0 — llmsFromPages() / llmsFullTxtFromPages() generate llms.txt and llms-full.txt from a page list; a first-class ## Optional block (doc.optional, round-trips through parseLlmsTxt); validateLlmsTxt(); and opt-in Markdown escaping via escapeLlmsText() / { escape: true }. All additive — existing output is byte-for-byte unchanged.

Install

npm install @lacspace/llms-txt      # or pnpm add / yarn add / bun add

Generate llms.txt

import { llmsTxt } from "@lacspace/llms-txt";

const txt = llmsTxt({
  title: "Lacspace",
  summary: "Open-source TypeScript packages and products.",
  details: "Zero-dependency, isomorphic, Lacspace-Free-Licensed.",
  sections: [
    {
      title: "Docs",
      links: [
        { title: "npm Packages", url: "https://lacspace.com/packages", notes: "20 packages" },
        { title: "SDK", url: "https://www.npmjs.com/package/@lacspace/sdk" },
      ],
    },
  ],
});

# Lacspace

> Open-source TypeScript packages and products.

Zero-dependency, isomorphic, Lacspace-Free-Licensed.

## Docs

- [npm Packages](https://lacspace.com/packages): 20 packages
- [SDK](https://www.npmjs.com/package/@lacspace/sdk)

Generate llms-full.txt

import { llmsFullTxt } from "@lacspace/llms-txt";

llmsFullTxt({
  title: "Lacspace Docs",
  sections: [
    { title: "Getting started", url: "https://lacspace.com/docs", content: "# Getting started\n\nInstall with npm…" },
  ],
});

Parse

import { parseLlmsTxt } from "@lacspace/llms-txt";

const doc = parseLlmsTxt(existing); // { title, summary, details, sections: [{ title, links }] }

Serve llmsTxt(...) at /llms.txt and llmsFullTxt(...) at /llms-full.txt (e.g. from a Next.js Route Handler or any static build step).

The Lacspace SEO Kit

| Package | For | | --- | --- | | @lacspace/seo | Metadata & JSON-LD | | @lacspace/sitemap | sitemap.xml | | @lacspace/robots | robots.txt | | @lacspace/llms-txt | llms.txt / llms-full.txt (this package) | | @lacspace/site-verify | Search-engine verification | | @lacspace/rss | RSS / Atom / JSON feeds | | @lacspace/slugify | SEO URL slugs |

New in 1.2 — build from your sitemap

import { llmsTxtFromSitemap } from "@lacspace/llms-txt";

// Feed the same URL list your sitemap uses; sections + titles are derived
const txt = llmsTxtFromSitemap(
  [
    { url: "https://x.com/", section: "Main" },
    { url: "https://x.com/docs/sdk", section: "Docs", title: "SDK" },
  ],
  { title: "Lacspace", summary: "Open-source packages & products." },
);

New in 1.1 — Response helpers

import { llmsTxtResponse, llmsFullTxtResponse } from "@lacspace/llms-txt";

// app/llms.txt/route.ts — served as text/plain
export function GET() {
  return llmsTxtResponse({ title: "Lacspace", summary: "…", sections });
}

Advanced (new)

All additive and backward-compatible — llmsTxt, llmsFullTxt, parseLlmsTxt, llmsTxtFromSitemap and the *Response helpers keep their existing behavior.

Build from routes

llmsTxtFromRoutes(routes, meta) turns a flat list of { title, url, notes?, section? } into an llms.txt, grouping by section (first-seen order preserved).

import { llmsTxtFromRoutes } from "@lacspace/llms-txt";

llmsTxtFromRoutes(
  [
    { title: "Home", url: "https://acme.com/", section: "Start" },
    { title: "API", url: "https://acme.com/api", notes: "reference", section: "Docs" },
    { title: "CLI", url: "https://acme.com/cli", section: "Docs" },
  ],
  { title: "Acme", summary: "Acme docs", defaultSection: "Docs" },
);

Sitemap: XML string, auto-sections, dedupe

llmsTxtFromSitemap now accepts a raw sitemap XML string as well as an array, de-duplicates repeated URLs, and can derive sections from the first path segment with sectionFromPath: true. The original array signature is unchanged.

llmsTxtFromSitemap(sitemapXmlString, { title: "Acme", sectionFromPath: true });
// "/docs/intro" → "## Docs", "/blog/hello" → "## Blog"

Sort links within sections

llmsTxt(doc, { sort }) — and the sort option on llmsTxtFromRoutes / llmsTxtFromSitemap — orders the links inside each section. Sections keep their array order.

llmsTxt(doc, { sort: "title" });        // ascending by title
llmsTxt(doc, { sort: "url-desc" });     // descending by url
llmsTxt(doc, { sort: (a, b) => /* custom */ 0 });

parseLlmsTxt round-trips what llmsTxt produces (title, summary, details, sections and links).

New in 1.4 — pages → both files, Optional block, validation

Build both files from a page list

llmsFromPages(pages, meta) returns { txt, full } — the compact llms.txt (a link per page with a url, grouped by section) and the expanded llms-full.txt (every page's content inlined). Also available split: llmsTxtFromPages / llmsFullTxtFromPages.

import { llmsFromPages } from "@lacspace/llms-txt";

const { txt, full } = llmsFromPages(
  [
    { title: "Home", url: "https://acme.com/", content: "# Home\n\nWelcome.", section: "Start" },
    { title: "API", url: "https://acme.com/api", content: "# API\n\n…", notes: "reference", section: "Docs" },
    { title: "Legacy", url: "https://acme.com/legacy", content: "# Legacy", optional: true },
  ],
  { title: "Acme", summary: "Acme docs" },
);
// serve `txt` at /llms.txt and `full` at /llms-full.txt

The ## Optional block

LlmsDoc.optional?: LlmsLink[] renders the special ## Optional section (links an LLM may skip) last, and round-trips through parseLlmsTxt. Pages with optional: true are routed there automatically.

Validate

import { validateLlmsTxt } from "@lacspace/llms-txt";

const { valid, issues } = validateLlmsTxt(text);
// valid === false only on errors (missing H1); warnings flag a missing
// summary, malformed link lines, duplicate/mis-placed H1.

Markdown escaping (opt-in)

escapeLlmsText(s) escapes \ [ ] ( ) in link text; pass { escape: true } to llmsTxt / the builders to apply it to titles and notes. Off by default — the default output is unchanged.

| Function | Returns | | --- | --- | | llmsTxt(doc, opts?) | llms.txt string (now supports doc.optional, opts.escape) | | llmsFullTxt(doc) | llms-full.txt string | | llmsFromPages(pages, meta) | { txt, full } from one page list | | llmsTxtFromPages(pages, meta) | llms.txt from pages | | llmsFullTxtFromPages(pages, meta) | llms-full.txt from pages | | llmsTxtFromRoutes(routes, meta) | llms.txt from flat routes | | llmsTxtFromSitemap(entries, meta) | llms.txt from a sitemap (array or XML) | | parseLlmsTxt(text) | { title, summary, details?, sections, optional? } | | validateLlmsTxt(text) | { valid, issues } | | escapeLlmsText(s) | escaped link text |

Licensing

This package is free under the Lacspace Free Licence — permissive freedoms. Use it in personal and commercial projects at no cost; just keep the notice.

Not every Lacspace package is free. We also offer Commercial (paid), Client-specific, and Private (proprietary) packages under separate terms. See the full Lacspace Licence Centre.

The Lacspace Developer Platform

@lacspace/llms-txt is part of 80+ zero-dependency, isomorphic TypeScript packages — one standard library for the modern web. Explore the ecosystem:

• 📦 This package, documented — https://developer.lacspace.com/packages/llms-txt
• 🗂️ All 80+ packages — https://developer.lacspace.com/packages
• 🧭 Developer handbook — guides & runnable recipes — https://developer.lacspace.com/handbook
• 🧪 Live playground — run any package in your browser — https://developer.lacspace.com/playground
• 🖥️ Finished app templates — https://templates.lacspace.com
• 🚀 Scaffold a full app — npm create lacspace-app@latest

Free under the Lacspace Free Licence — a permissive, free-to-use licence.