# Fix llms.txt: publish a short Markdown guide at /llms.txt > An llms.txt file is a short Markdown file at the root of your site that tells language models and agents what the site is and which pages matter. Agents read it on demand, while answering a question about you, which is why it is worth twenty minutes. Checked 2026-09-30 against Agent-Readiness Grade 1.3.0. HTML version: https://grade.agentexchange.work/fix/llms-txt ## What the grade checks - `GET https://example.com/llms.txt` (and `/ai.txt`) with a 7-second timeout. It counts when the answer is HTTP 200, text rather than an HTML page from a catch-all route, and not empty. - Either file earns the llms.txt point of the machine-readable surfaces area. The area adds one point for a sitemap and one for an agent surface (agent-card.json, openapi.json, mcp-registry-auth or x402), capped at 2. 1 of the 2 points in the machine-readable surfaces area. ## Why it matters for AI agents and crawlers The [llms.txt proposal, v2](https://llmstxt.org/) defines the file: an H1 with the site or project name is the only required part, followed by a blockquote summary, optional notes, and H2 sections that list links as `[name](url): notes`. A section named `Optional` marks links an agent can skip when it is short of context. It is past the experiment stage: the proposal notes that Chrome's Lighthouse [audits sites for one](https://developer.chrome.com/docs/lighthouse/agentic-browsing/llms-txt) in its agentic-browsing checks, that documentation platforms generate it, and that OpenAI, Anthropic and Google publish llms.txt files for their own developer docs. It is not a crawler control. Access is still decided by robots.txt; llms.txt guides an agent that is already allowed in. Point its links at pages that answer questions (pricing, docs, product pages, policies). The v2 proposal also recommends clean Markdown copies of pages (`page.md` or `page.html.md`) and a `Link: ; rel="describedby"` header where you can add one. ## How to fix it ### 1. Write the file Start with the name and one sentence a stranger would understand, then list the pages that answer the questions people ask about you. Keep it short enough to read in one go: the detail lives behind the links. The [llms.txt generator](https://grade.agentexchange.work/tools/llms-txt-generator) drafts one from your home page and sitemap. llms.txt: ```markdown # Example Co > Example Co makes invoicing software for small agencies: send invoices, track payments and chase late clients automatically. Plans from $12 a month. Example Co is a web app; there is no desktop client. Support answers within one business day. ## Product - [Pricing](https://example.com/pricing): plans, limits and what each includes - [Features](https://example.com/features): invoicing, reminders, payment links - [Integrations](https://example.com/integrations): Stripe, QuickBooks, Xero ## Docs - [Getting started](https://example.com/docs/start.md): create an account and send the first invoice - [API reference](https://example.com/docs/api.md): REST endpoints and authentication ## Optional - [Blog](https://example.com/blog): product updates and guides - [Privacy policy](https://example.com/privacy) ``` ### 2. Publish it at /llms.txt Serve it at the root as `text/plain` (or `text/markdown`). What matters is that the answer is the file, not an HTML page. #### Static site Save it as `llms.txt` in the web root. Web servers send `.txt` files as `text/plain`. #### WordPress [Yoast SEO](https://yoast.com/features/llms-txt/) and [AIOSEO](https://aioseo.com/features/llms-txt/) can generate and maintain the file. To write it by hand, upload `llms.txt` to the WordPress root folder (the one with `wp-config.php`) by SFTP or your host's file manager: the web server returns a real file before WordPress routing runs. #### Next.js Put it at `public/llms.txt`, or build it in a [route handler](https://nextjs.org/docs/app/api-reference/file-conventions/route) so it stays in sync with your content: app/llms.txt/route.ts: ```ts export async function GET() { const body = `# Example Co > Invoicing software for small agencies. ## Product - [Pricing](https://example.com/pricing): plans and limits ` return new Response(body, { headers: { 'Content-Type': 'text/plain; charset=utf-8' }, }) } ``` #### Cloudflare With [Workers static assets](https://developers.cloudflare.com/workers/static-assets/) or Pages, put `llms.txt` in the directory you deploy (often `public/`). From Worker code, return it with a text content type: Cloudflare Worker: ```js const LLMS = `# Example Co > Invoicing software for small agencies. `; export default { async fetch(request, env) { if (new URL(request.url).pathname === "/llms.txt") { return new Response(LLMS, { headers: { "content-type": "text/plain; charset=utf-8" } }); } return env.ASSETS.fetch(request); // your static assets binding, if the Worker has one }, }; ``` ### 3. Point agents to it (optional) The v2 proposal recommends the `describedby` link relation so a client can find the file from any page, as an HTML tag or an HTTP header. HTML head, or response header: ```html Link: ; rel="describedby" ``` ## How to verify Check the status and the content type, then read the first lines. ```sh curl -s -o /dev/null -w "%{http_code} %{content_type}\n" https://example.com/llms.txt curl -s https://example.com/llms.txt | head -5 ``` `200 text/plain` and a first line starting with `# `. If the content type is `text/html`, a catch-all route is answering and the grade will not count it. Re-grade: https://grade.agentexchange.work/grade?url=example.com&fresh=1 ## Questions ### Is llms.txt an official standard? It is a public proposal by Jeremy Howard of Answer.AI, dated 2024-09-03 and revised as v2, not an IETF or W3C standard. It is widely used in practice: Lighthouse audits for it and documentation platforms generate it. ### Does llms.txt replace robots.txt or sitemap.xml? No. robots.txt says what crawlers may access, sitemap.xml lists every indexable page, and llms.txt is a short, curated guide an agent reads when it needs to understand the site. The proposal describes them as complementary. ### What goes in the Optional section? By convention, links an agent can skip when it needs a shorter context: blogs, changelogs, long references, policies. ### Can I generate it automatically? Yes. The llms.txt generator on this site drafts one from your home page and sitemap, and WordPress plugins (Yoast SEO, AIOSEO) and docs platforms (Mintlify, GitBook) generate and maintain one. Edit the summary line by hand either way: it is the line agents quote. ## Sources - [The /llms.txt file, v2](https://llmstxt.org/) (llmstxt.org (Jeremy Howard, Answer.AI)) - [Lighthouse: llms.txt audit (agentic browsing)](https://developer.chrome.com/docs/lighthouse/agentic-browsing/llms-txt) (Chrome for Developers) - [Yoast SEO: llms.txt](https://yoast.com/features/llms-txt/) (Yoast) - [AIOSEO: llms.txt generator](https://aioseo.com/features/llms-txt/) (AIOSEO) - [Route handlers (route.ts), including non-UI responses](https://nextjs.org/docs/app/api-reference/file-conventions/route) (Next.js) - [Workers static assets](https://developers.cloudflare.com/workers/static-assets/) (Cloudflare) ## Related - [robots.txt for AI crawlers](https://grade.agentexchange.work/fix/robots-txt-ai-crawlers.md): robots.txt groups for eight AI crawler tokens; a blanket Disallow: / counts as blocked. - [sitemap.xml](https://grade.agentexchange.work/fix/sitemap-xml.md): An XML sitemap at /sitemap.xml, or a Sitemap: line in robots.txt pointing anywhere. - [ai.txt](https://grade.agentexchange.work/fix/ai-txt.md): An alternative to llms.txt for the same point; no crawler vendor documents reading it. - [llms.txt generator](https://grade.agentexchange.work/tools/llms-txt-generator) - [All fix guides](https://grade.agentexchange.work/fix)