llms.txt Pro Skill
getcited/llmstxt-pro-skill
Gere llms.txt spec-compliant para 26 CMS e 10 agentes de IA direto no Claude
Ver no GitHubSkill para Claude Code que gera e deploya arquivos llms.txt compatíveis com especificação em 26 plataformas CMS. Otimiza a descoberta de seu site por agentes de IA como ChatGPT, Claude, Gemini e Perplexity através de um manifesto padronizado no raiz do domínio.
Instalação
npx convez-ai getcited/llmstxt-pro-skill1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
Casos de uso
- Preparar sites para serem descobertos e citados por agentes de IA de busca
- Implementar llms.txt em WordPress, Shopify, Webflow, Wix e outras 21 plataformas
- Documentar estrutura e conteúdo do site para ChatGPT, Claude, Gemini, Perplexity e Copilot
- Entregar arquivo llms.txt otimizado para agências que gerenciam múltiplos clientes
Veredito

Skill bem estruturado que preenche lacuna real no ecossistema — é a única implementação como Claude Code skill com suporte robusto a 26 plataformas. Útil para quem quer garantir citabilidade em busca por IA sem abandonar o contexto do projeto.
README
A tradução para o português deste README ainda está na fila.
llms.txt Pro
The only Claude Code skill for llms.txt. Generate and deploy spec-compliant llms.txt files on any platform — from WordPress and Shopify to headless CMS setups and anything in between.
llms.txt is the AI-era equivalent of robots.txt. It is a plain text markdown file placed at your domain root that tells AI search agents — ChatGPT, Claude, Gemini, Perplexity, Copilot, and others — what your site is about, what it contains, and how to interpret it. As AI tools increasingly answer searches before users click, a well-formed llms.txt is one of the clearest signals you can give to get cited.
Why this skill
Every other llms.txt tool is a web app, standalone CLI, or CMS plugin. This is the only Claude Code skill — meaning it lives in your project, knows your context, and generates platform-appropriate files without copy-pasting between tools.
What this skill covers
- The full llmstxt.org format specification with every rule documented
- Content selection and quality rules — not just format compliance, but what makes a file strong enough to drive citations
- Platform-by-platform implementation for 26 CMS and hosting platforms
- Cloudflare Workers fallback — works for any domain, regardless of platform
- Which AI agents consume
llms.txtand how each one uses it - A verification checklist covering file delivery, spec compliance, content quality, and URL health
- Client deliverable format for agencies
AI agents that consume llms.txt
| Agent | Network |
|---|---|
| ChatGPT | OpenAI / Bing |
| Claude | Anthropic |
| Gemini | |
| Copilot | Microsoft / Bing |
| Perplexity | Independent |
| Grok | xAI |
| You.com | You.com |
| Brave Leo | Brave |
| Kagi Assistant | Kagi |
| Phind | Phind |
See docs/ai-agents.md for how each agent uses the file and how to measure citation impact.
CMS platform coverage
| Platform | Method | Automated updates | Guide |
|---|---|---|---|
| WordPress + Yoast SEO Free | Toggle in plugin settings | Weekly | docs/platform-wordpress.md |
| WordPress (manual) | File upload to root | Monthly | docs/platform-wordpress.md |
| Shopify | Page redirect or Hydrogen route | Manual | docs/platform-shopify.md |
| Webflow | Cloudflare Worker | Manual | docs/platform-webflow.md |
| Wix | Native toggle (rolling out) or Cloudflare Worker | Auto / Manual | docs/platform-wix.md |
| Squarespace | Cloudflare Worker | Manual | docs/platform-squarespace.md |
| Ghost (self-hosted) | File in web root | Manual | docs/platform-ghost.md |
| Ghost Pro | Cloudflare Worker | Manual | docs/platform-ghost.md |
| Framer | Cloudflare Worker | Manual | docs/platform-framer.md |
| HubSpot CMS | Cloudflare Worker or page redirect | Manual | docs/platform-hubspot.md |
| TYPO3 | Native extension (ai-llms-txt) | Auto on cache clear | docs/platform-typo3.md |
| Drupal | Native module (llms_txt) | Auto on content update | docs/platform-drupal.md |
| Next.js | Route handler (app/llms.txt/route.js) | On deploy | docs/platform-headless-cms.md |
| Nuxt | Server route or Nuxt Content module (native) | On deploy | docs/platform-headless-cms.md |
| Astro | Page route (src/pages/llms.txt.js) | On deploy | docs/platform-headless-cms.md |
| SvelteKit | Server route (src/routes/llms.txt/+server.js) | On deploy | docs/platform-headless-cms.md |
| Prismic + any frontend | Dynamic API fetch via @prismicio/client | On deploy | docs/platform-headless-cms.md |
| Payload CMS | Custom root endpoint in payload.config.ts | On deploy | docs/platform-headless-cms.md |
| Contentful + any frontend | CMS-driven route handler | On deploy | docs/platform-headless-cms.md |
| Strapi + any frontend | CMS-driven route handler | On deploy | docs/platform-headless-cms.md |
| Sanity + any frontend | CMS-driven route handler | On deploy | docs/platform-headless-cms.md |
| VitePress | docs/public/llms.txt or vitepress-plugin-llms | On build | docs/platform-static.md |
| Plain HTML | File in root folder | Manual | docs/platform-static.md |
| Jekyll | File in project root | On build | docs/platform-static.md |
| Hugo | static/llms.txt | On build | docs/platform-static.md |
| Eleventy | File in output folder | On build | docs/platform-static.md |
| Any other CMS | Cloudflare Worker | Manual | docs/platform-webflow.md |
Installation
Copy the skill file to your project's .claude/skills/ directory:
cp .claude/skills/llmstxt-pro.md /your-project/.claude/skills/llmstxt-pro.md
Global install — available across all projects:
cp .claude/skills/llmstxt-pro.md ~/.claude/skills/llmstxt-pro.md
Then add llmstxt-pro to your Claude Code agent's installed skills list. The skill activates automatically when you ask Claude Code to create or deploy an llms.txt file.
How it works
┌─────────────────────────────────────────────────────┐
│ 1. YOU ASK │
│ "Create an llms.txt for my site" │
└─────────────────────────────────────────────────────┘
│
▼
┌─────────────────────────────────────────────────────┐
│ 2. SKILL READS YOUR CONTEXT │
│ Platform detection (WordPress/Shopify/etc.) │
│ Site name, top pages, recent posts │
│ CMS capabilities and constraints │
└─────────────────────────────────────────────────────┘
│
▼
┌─────────────────────────────────────────────────────┐
│ 3. FILE GENERATED │
│ Spec-compliant format (H1, blockquote, sections)│
│ Content selection and quality rules applied │
│ Colon descriptions on every post entry │
└─────────────────────────────────────────────────────┘
│
▼
┌─────────────────────────────────────────────────────┐
│ 4. DEPLOYMENT INSTRUCTIONS │
│ Platform-specific steps │
│ Verification checklist │
│ Maintenance schedule │
└─────────────────────────────────────────────────────┘
Format specification
The file follows the llmstxt.org spec. Sections appear in this fixed order:
# Site Name ← REQUIRED. H1. Site or project name only.
> One-sentence summary of the site. ← OPTIONAL. Blockquote. Most important for AI citation matching.
Paragraph detail about the site. ← OPTIONAL. Prose. Entity-dense, specific, no marketing fluff.
## Section Name ← OPTIONAL. H2. Groups a set of related links.
- [Page Title](https://url): Note ← Link required. Colon note optional but strongly recommended.
## Optional ← SPECIAL. AI agents may skip this section in short contexts.
- [Secondary resource](https://url)
The three rules that matter most:
- H1 is the first line. One H1 only. Site name only — nothing else.
- Every list item must contain a markdown hyperlink. Bare URLs are invalid.
- Colon descriptions on post entries are how agents learn what a page covers without fetching it.
Content quality
Format compliance is the floor. See docs/content-quality.md for what separates a file that passes validation from one that actually drives AI citations — covering blockquote strength, colon description quality, page selection logic, freshness signals, and common mistakes.
Reference example
examples/example-yoast-output.txt shows what a Yoast-generated file looks like in practice. Use it as the quality benchmark when writing files manually.
The Yoast output is spec-valid but minimal. It omits the blockquote and descriptive prose. A manually crafted file should include both — they give AI agents enough context to cite the site without fetching any pages.
Verification checklist
After deploying any llms.txt file, run the full checklist at docs/verification-checklist.md. Key checks:
- File resolves at
https://yourdomain.com/llms.txtwith no redirect chain - First line is H1 with site name only
- Every list item has a working hyperlink
- No URLs return 404
- No noindexed pages listed
- Colon descriptions on all post entries
- Content is current — posts within the last 12 months
Repository structure
llmstxt-pro-skill/
├── .claude/skills/llmstxt-pro.md ← Install this file in your project
├── examples/
│ ├── example-yoast-output.txt ← What a Yoast-generated file looks like
│ └── template-manual.txt ← Blank template for manual files
├── docs/
│ ├── ai-agents.md ← Which AI agents consume llms.txt and how
│ ├── content-quality.md ← What makes a file strong enough to drive citations
│ ├── platform-wordpress.md
│ ├── platform-shopify.md
│ ├── platform-webflow.md
│ ├── platform-wix.md
│ ├── platform-squarespace.md
│ ├── platform-ghost.md
│ ├── platform-framer.md
│ ├── platform-hubspot.md
│ ├── platform-typo3.md
│ ├── platform-drupal.md
│ ├── platform-headless-cms.md ← Next.js, Nuxt, Astro, SvelteKit, Prismic, Payload, Contentful, Strapi, Sanity
│ ├── platform-static.md ← HTML, Jekyll, Hugo, Eleventy, VitePress
│ └── verification-checklist.md
├── skill.json
├── CLAUDE.md
└── LICENSE
License
MIT — Nil Boraquit