Ready for AI agents.
AI agents browse your site, compare plans and take action. Here is what makes a site ready for them, and what Paradigma checks.
What makes a site agent-ready. Every item is checked on your pull requests or your live site.
- Search and AI crawlers allowedLive
- robots.txt lets the crawlers that power search and AI answers read your pages, while you stay free to opt out of training.AI search crawlers blockedSearch engines blocked in robots.txtCrawlers excluded by the live robots.txt
- Clear titles, descriptions and canonicalsLive
- Every public page says what it is and where it lives, so an agent can cite the right URL.Page title or description removedCanonical URL changed or removedRendered pages without a title
- Structured dataLive
- Products, prices, articles and FAQs stated as facts in JSON-LD, not only in prose.Structured data removedInvalid structured data
- Pages that answer with the right statusLive
- A deployed page that errors is caught before it merges.Pages that error on the deployment
- Readable without JavaScriptLive
- The main content is in the HTML the server sends, not loaded later by scripts. Paradigma flags pages that become client-only in a pull request, live pages that are an empty JavaScript shell, and live pages that show much more text or links once scripts run.Pages that load their content only in the browserLive pages that are empty without JavaScriptLive pages whose content needs JavaScriptContent lost on the PreviewLinks that need a click to work
- Page languageLive
- A declared
<html lang>, so agents, screen readers and search engines read and quote each page in the right language.Live pages without a language - llms.txt, if you publish oneLive
- A Markdown file at /llms.txt for agents and developers working with your product. Paradigma checks it follows the format Lighthouse audits and that its links work. Google Search doesn't use it, so it is not a way to get cited, and not having one is fine.llms.txt that fails the format checkllms.txt that returns a web pagellms.txt links that return errors
- Agent catalog (ard.json)Live
- A machine-readable list of your APIs and MCP servers at /.well-known/ard.json, following the Agentic Resource Discovery proposal that Lighthouse audits.Agent catalog (ard.json) that doesn't follow the formatAgent catalog only at the old pathard.json entries with too few or too many example queries
- Forms agents can use (WebMCP)Live
- Forms that declare
toolnameandtooldescription, so browser agents can search, book or ask without guessing at the layout. Optional while Chrome runs it as a trial.Forms that AI agents can't use as tools
Two small files, big difference. Illustrative examples for a Next.js app.
app/layout.tsxBefore
export default function RootLayout({ children }) {
return (
<html>
<body>{children}</body>
</html>
);
}After
export default function RootLayout({ children }) {
return (
<html lang="en">
<body>{children}</body>
</html>
);
}A declared language helps agents and assistive technology read the page correctly.
public/llms.txtBefore
(no file: /llms.txt returns 404)After
# Acme
> Acme is invoicing for freelancers: send an invoice
> in a minute and get paid by card or bank transfer.
## Pages
- [Pricing](https://example.com/pricing): plans and limits
- [Docs](https://example.com/docs): how to get startedQuestions about agent readiness
- What is an agent-ready website?
- One that AI agents can read, navigate and act on: the content is in the HTML, pages say clearly what they are, crawlers are allowed, and machine-readable files like llms.txt point agents to what matters.
- Is agent readiness the same as GEO?
- They overlap. GEO is about being mentioned and cited in AI answers; agent readiness is about AI agents being able to use your site once they arrive. Both start with pages crawlers can read.
- What is llms.txt?
- A short Markdown file at /llms.txt that summarizes your site and links to the pages that matter. Agents and coding assistants read it when they work with your product; Google Search does not use it, so it won't get you cited on its own.
More terms in the glossary, and how this fits with getting cited in the GEO guide.