TL;DR
- 780+ Verified Implementations: Over 780 live websites across AI research, developer tools, cloud infrastructure, and SaaS now serve llms.txt files.
- Industry Leaders Leading Adoption: Tech giants and documentation leaders including Cloudflare, Anthropic, Supabase, Vercel, and ElevenLabs provide curated Markdown paths for coding agents and LLM crawlers.
- Platform Automation: Platforms like Mintlify, GitBook, and Wix now automatically generate llms.txt and llms-full.txt files out-of-the-box.
- Searchable Live Directory: Browse the complete database of live implementations on the llms.txt Directory.
Generate your llms.txt file
Paste your domain and get a structured index file for ChatGPT, Claude & Perplexity instantly.
As of 2026, over 780 live websites use llms.txt files to guide AI search engines and autonomous coding agents through their technical content architectures. Major infrastructure providers, including Anthropic, Cloudflare, Supabase, and Vercel, utilize these Markdown directories to ensure LLMs can cite their documentation with high precision, dramatically reducing hallucinations in AI-generated code.
Comprehensive 2026 Adoption Overview
The adoption of the specification has matured significantly since its inception in late 2024. Analysis of the verified llms.txt directory reveals that implementation is heavily skewed toward technical organizations that rely on developer experience and API documentation. Real-time usage spans four primary industry verticals:
- AI and LLM Platforms: Foundational model providers (Anthropic, ElevenLabs) who have a direct stake in establishing deterministic agentic protocols.
- Developer Infrastructure & Databases: Cloud hosting providers and database engines (Cloudflare, Supabase, Vercel, Firecrawl) where accurate code generation in IDEs is a core product differentiator.
- SaaS & Enterprise Software: B2B companies using Business-to-Agent (B2A) routing to point AI assistants directly to verified pricing, security, and integration docs.
- Open Source Projects & Frameworks: Framework ecosystems ensuring AI tools like Cursor and Claude Code generate up-to-date syntax rather than hallucinated legacy code.
Detailed Case Studies of Live Implementations
By evaluating live llms.txt examples, several industry leaders demonstrate how to engineer these files for maximum agentic impact:
1. Anthropic
As a pioneer in large language models, Anthropic utilizes the specification to map documentation for its Claude API. Anthropic publishes both an /llms.txt file for rapid routing and an extensive /llms-full.txt file that concatenates its entire API reference. This dual-file strategy allows developers utilizing Claude Desktop or Cursor to instantly inject authoritative, up-to-date Claude API limits and endpoints directly into their local context window.
2. Cloudflare
Operating massive volumes of edge computing documentation, Cloudflare leverages the file to differentiate between its varied product lines, ranging from Workers and Pages to D1 and Zero Trust architectures. The Cloudflare llms.txt is segmented meticulously via level-2 headers, ensuring that an autonomous agent answering a query about the "Wrangler CLI" does not ingest irrelevant context regarding DDoS mitigation protocols.
3. Supabase
Supabase utilizes the standard to optimize developer experience in IDEs like Cursor and Windsurf. By pointing agents to clean Markdown versions of its PostgreSQL and edge function tutorials, the Supabase llms.txt ensures that when a developer prompts their editor to construct an authentication flow, the generated code reflects the latest SDK syntax rather than outdated boilerplate. Furthermore, Supabase and similar platforms deploy an llms-small.txt variant, offering a curated low-token subset of documentation specifically tuned for smaller context windows.
4. Vercel
For Next.js and Vercel infrastructure deployments, the documentation is notoriously complex and frequently updated. Vercel implements a dynamic route handler for its llms.txt file, ensuring the file automatically updates in sync with new framework releases. This dynamic implementation guarantees that AI agents fetching Next.js App Router rules receive the absolute latest specifications without requiring manual file uploads.
5. Firecrawl
As a tool designed specifically for scraping and transforming web content into LLM-ready Markdown, Firecrawl natively supports and generates llms.txt files for its clients. Their own corporate implementation serves as a best-in-class example of recursive AI optimization, documenting their mapping parameters and API endpoints in a heavily token-optimized format.
6. ElevenLabs
The voice AI platform relies on the file to streamline access to its REST API and WebSocket protocols. Because audio streaming parameters are highly technical and heavily dependent on correct JSON structuring, ElevenLabs' implementation focuses on routing coding agents directly to integration templates, intentionally bypassing marketing and pricing pages to preserve context limits for technical execution.
7. Mintlify
Mintlify represents a structural shift in the ecosystem, as the documentation platform now automatically generates both llms.txt and llms-full.txt files for every single documentation site it hosts. This platform-level integration is responsible for a massive percentage of verified implementations globally, automating the compliance layer for thousands of companies.
8. Cursor
Cursor is uniquely positioned as both a publisher and a primary consumer of the standard. As an AI-native IDE, Cursor's internal @Docs feature actively seeks out implementations across the web to pull in remote context. Concurrently, Cursor's own corporate website uses the file to instruct external models on how to utilize new IDE features, demonstrating a seamless, closed-loop B2A ecosystem.
Enterprise Implementations Comparison
To understand the architectural variations in implementation, the following table compares key metrics across verified directory examples:
| Organization | Primary File Hosted | Companion File Hosted | Est. Token Count (Full) | Primary Agent Targets |
|---|---|---|---|---|
| Anthropic | /llms.txt |
/llms-full.txt |
~125,000 | Claude Code, Cursor |
| Vercel | /llms.txt |
N/A | N/A | GitHub Copilot, Windsurf |
| Mintlify (Platform) | /llms.txt |
/llms-full.txt |
Variable | IDE Agents, MCP Servers |
| Supabase | /llms.txt |
/llms-small.txt |
~85,000 | Cursor, LLM Crawlers |
| Aptos | /llms.txt |
/llms-full.txt |
~140,000 | Claude.ai, GitHub Copilot |
Explore and Submit to the Directory
Want to see how other companies in your industry structure their files? Explore hundreds of verified examples or submit your site to the live database in the llms.txt Directory. You can also generate your own standard-compliant file in minutes with the free llms.txt generator.
Frequently Asked Questions
Frequently Asked Questions
How do I view the llms.txt file of a specific website?
To view any site's discovery file, append /llms.txt directly to its canonical domain in any web browser. You can also inspect raw headers and Markdown payload in your terminal using curl -I https://example.com/llms.txt. If a site maintains a full-context companion file, it can similarly be accessed at /llms-full.txt.
Are verified llms.txt implementations mostly static or dynamic?
While early adopters created static text files manually, modern enterprise platforms use dynamic route handlers that query CMS backends. Frameworks like Next.js, Astro, and Remix generate the Markdown payload dynamically with edge caching. This guarantees the machine-readable index stays continuously synchronized with live documentation.
Why do some sites host an llms-full.txt file instead of just llms.txt?
The primary llms.txt file serves as a routing directory that directs AI crawlers to specific individual URLs. In contrast, llms-full.txt bundles the entire markdown documentation into a single, cohesive file. Advanced reasoning models with large context windows ingest this full payload in a single HTTP fetch, bypassing multi-step crawl latency.
Do e-commerce websites use the llms.txt specification?
E-commerce is now the largest adopter of llms.txt following Shopify's platform-wide release of native /agents.md and /llms.txt endpoints across all Online Stores. Merchants can override these endpoints using custom Liquid templates (templates/llms.txt.liquid) to expose curated product categories, policies, and shopping assistant prompts. This allows AI shopping agents in ChatGPT and Claude to cite accurate inventory and pricing without scraping errors.
Can I see live llms.txt examples on platforms other than custom codebases?
Yes, major CMS platforms, documentation engines, and e-commerce hosts now integrate llms.txt natively out of the box. Hosted platforms including Shopify, Wix, GitBook, Mintlify, and Duda compile compliant files automatically for all published websites. You can explore hundreds of verified, real-world examples across our live directory at llms-text.com/directory.
Ready to index your website for ChatGPT, Claude & Perplexity?
Generate your standardized llms.txt and full GEO context pack in seconds.