Generate both the index (llms.txt) and the structured database template (llms-full.txt) side-by-side for your website root.
Input your website details to export standardized, AI-ready indexing files.
Creating an llms.txt index is the foundational step toward Generative Engine Optimization (GEO). Traditional search bots read visual URLs via XML sitemaps, but generative models require text-centric directories to build summaries in search query answers. Our Generator tool abstracts away the complexity of markdown syntax, allowing you to quickly export standard-compliant indexes and database templates for your website root.
Large Language Models (LLMs) do not process the web like human browsers. They do not care about your CSS frameworks, your JavaScript bundle sizes, or your responsive flexbox layouts. They ingest tokens—chunks of raw text.
When an AI agent (such as a RAG pipeline or a custom AI crawler) arrives at your site, its goal is to extract maximum semantic value with minimum token expenditure. A poorly formatted file forces the model to guess the hierarchy of your documentation, often leading to hallucinations in the answers it provides to end-users.
| Element | Traditional SEO | Generative Engine Optimization (GEO) |
|---|---|---|
| Discovery | sitemap.xml |
llms.txt (Table of Contents) |
| Content Delivery | HTML Pages scattered across URLs | llms-full.txt (Unified Knowledge Base) |
| Ranking Factor | Backlinks and keyword density | Semantic density and factual accuracy |
| Output | A list of 10 blue links | A conversational, synthesized answer |
Read our full analysis on the SEO benefits of adopting llms.txt early.
Our generator strictly adheres to the official Answer.ai specification. Here is a breakdown of the structural components it builds for you:
llms.txt and llms-full.txtOur generator produces two distinct outputs. Understanding when to use which is critical for optimization.
llms.txt)This is your routing table. It should rarely exceed 1,500 words. When a crawler hits your domain, it reads this first. If a user asks an AI, "How do I authenticate with Acme Corp?", the AI agent will fetch your llms.txt, see the link to /docs/auth, and execute a targeted fetch for just that page. It preserves the model's context window.
llms-full.txt)This is the heavy lifter. If a developer uses a tool like Cursor or GitHub Copilot and adds your entire site to their workspace via `@https://acme.com/llms.txt`, the IDE will typically look for the `llms-full.txt` reference. This file contains the concatenated markdown of your entire documentation suite. By providing this directly, you prevent the IDE from having to spider 50 different URLs, vastly speeding up the developer experience.
Pro Tip: Generating this manually is tedious for large sites. For enterprise-scale applications, you should look into generating llms-full.txt programmatically via CI/CD pipelines.
When using the "Resources & Links Mapping" section of our generator, the text you choose for the hyperlink (the anchor text) is incredibly important. Unlike human users who might infer meaning from visual context, LLMs rely entirely on semantics.
[Click Here](https://acme.com/pricing) (Zero semantic value)[Pricing](https://acme.com/pricing) (Okay, but vague)[Enterprise Pricing and Tier Limits](https://acme.com/pricing) (Highly descriptive, perfect for RAG retrieval)No. The generator provides a visual UI where you input standard text fields. It handles the markdown formatting, escaping, and structuring behind the scenes to ensure 100% compliance with AI parsing specifications.
Yes, but you should only include public-facing URLs. Do not include links to private dashboards, admin panels, or authenticated API endpoints. The file is completely public and will be ingested into LLM training sets.
Both llms.txt and llms-full.txt must be placed in the root directory of your web server (e.g., https://yourdomain.com/llms.txt). This is where automated agents are hardcoded to look for them.
A sitemap is XML designed for traditional search engine crawlers (Googlebot). It contains metadata like lastmod and changefreq. llms.txt is markdown designed for language models to understand the semantic hierarchy and context of your site. You need both.
The standard strictly requires absolute URLs (starting with https://). AI agents often fetch your file outside the context of a browser frame. If they encounter a relative URL (like /docs/api), they will not know which domain to append it to, and the link will fail to resolve.
Our generator focuses on the structural index, which primarily consists of headers and lists. While markdown technically supports images and tables, including them in the main llms.txt is discouraged as it bloats the token count. Save those for the llms-full.txt.
You can edit the raw .txt files directly on your server using any text editor, or you can return to this generator, input your new links, and replace the old files. Always run your updated files through our validator before publishing.
While there is no hard technical limit, brevity is key in the index file. We recommend keeping the detailed description under 500 characters. If you have extensive architectural documentation, link out to it rather than pasting it into the index.
Absolutely. For e-commerce, use the generator to link to your highest-level category pages, shipping policies, and return documentation. Do not link to 10,000 individual product SKUs—that is what your XML sitemap is for.
Yes. If you host your docs at docs.example.com and your main site at example.com, you should generate two separate llms.txt files and host them at the root of their respective subdomains.