The AI Agent-Readiness Checklist: 30 Checks to Make Your Website Usable by AI in 2026

✅ Checklist · Updated October 2026 · 6 min read · by Jacob Campbell

"AI-ready" has become a marketing phrase, so here is what it means in practice. A website is agent-ready when an AI assistant or agent can find it, is allowed to read it, can understand it without guessing, and can act on it where that makes sense. Those four verbs map onto six layers, and each layer depends on the one before it. An MCP server on a site that blocks AI crawlers and is not indexed by Bing is a front door on a house with no road. This checklist walks the layers in order, with a concrete test for every item, and links to our in-depth guide for each.

Key facts

• 6 layers — Access, indexing, structure, text, discovery, action
• 30 checks — Each with a pass/fail test you can run yourself
• /100 — WebAgentScan scores the same ground automatically
• Order matters — Each layer depends on the one before it

Layer 1: access (can agents reach you?)

1. robots.txt returns 200 and parses. Test: open /robots.txt.
2. AI search and user-triggered bots are allowed (for example OAI-SearchBot, ChatGPT-User, Claude-SearchBot, Claude-User, PerplexityBot), even if you block training bots. See robots.txt for AI crawlers.
3. Your CDN or firewall is not silently blocking them. Cloudflare and others offer AI-bot blocking toggles and managed robots.txt that can override yours. Test: check your CDN’s bot settings and the robots.txt it actually serves.
4. Pages render without JavaScript. Test: curl a page and look for your main content in the HTML. Many AI fetchers do not run scripts.
5. No interstitials, cookie walls or logins in front of public content.

Layer 2: indexing (are you in the indexes assistants use?)

6. Indexed in Google: Google Search Console shows your key pages as indexed. Google’s AI features draw on Google’s index.
7. Indexed in Bing: Bing Webmaster Tools shows them too. Several assistants’ search features have drawn on Bing results, so this is not optional.
8. Sitemap submitted to both, with accurate lastmod dates. See where to submit your sitemap.
9. IndexNow enabled, so Bing, Yandex, Seznam, Naver and Yep hear about changes the same day. See the IndexNow guide.
10. No accidental noindex on pages you want found. Test: search the HTML and headers for noindex.

Layer 3: structure (can they understand you?)

11. Organization or LocalBusiness JSON-LD on the home page, with name, URL, logo, contact, address and sameAs links. See local business schema for AI search.
12. BreadcrumbList on inner pages, matching the visible breadcrumb.
13. FAQPage, Product, Service or Article markup where the content really is that type.
14. One clear H1 per page and a logical heading hierarchy.
15. Facts in text, not images: prices, hours, areas and phone numbers readable as text.
16. Consistent name, address and phone everywhere: site, Google Business Profile, directories.

Layer 4: text (can they read you cheaply?)

17. /llms.txt exists, follows the format and links to live pages. See what is llms.txt.
18. /llms-full.txt exists with the full text of key pages.
19. Markdown versions of pages via Accept: text/markdown and/or .md twins. See Markdown for AI agents.
20. Vary: Accept and canonical headers on negotiated responses.
21. rel=alternate links to the Markdown version in each page head.

Layer 5: discovery (can they find your interfaces?)

22. /.well-known/security.txt (RFC 9116).
23. An agent card at /.well-known/agent-card.json if you offer agent interfaces.
24. An MCP server card describing your MCP endpoint.
25. OAuth protected-resource metadata (RFC 9728) for any protected API or MCP server.
26. agents.txt or llms.txt lines pointing at all of the above. See agent discovery files.

Layer 6: action (can they do something?)

27. An MCP server on Streamable HTTP, exposing read-only lookups (services, hours, area) and, if you want leads, one write tool. See adding an MCP server.
28. Tool annotations and output schemas on every tool.
29. Rate limits, input validation and Origin checks on the endpoint.
30. Logging and monitoring: you can see which agents call which tools, and you check monthly.

How to run the checklist in 10 minutes

Run WebAgentScan on your home page. It scores most of layers 1, 3, 4, 5 and 6 out of 100 and lists what failed. Then open Google Search Console and Bing Webmaster Tools for layer 2. That covers 27 of the 30 checks; the rest (CDN settings, NAP consistency, monitoring) are a quick manual look.

What results to expect, honestly

Agent readiness removes reasons for AI systems to skip you. It does not create demand, and it does not override authority. When we made this site fully agent-ready we measured what changed: more than 200,000 IndexNow submissions, a 28,000-URL sitemap, and Bing indexing that moved far more slowly than the submissions, with referring domains (16 at the time) as the real constraint. The full numbers are in our agent-readiness experiment. The lesson: do the checklist, then spend the time on content and links that make you worth citing. See how to get cited by AI assistants.

Getting the technical layers done

• DIY: every item above is documented in the linked guides. Budget a few days for layers 4 to 6 if you are comfortable with PHP and .htaccess. Our pack vs DIY breakdown estimates it item by item.
• Any PHP site: the AI Indexing Pack covers layers 4, 5 and 6 plus bot access control, in about 30 minutes.
• WordPress: the WordPress edition does the same as a plugin. See also making WordPress readable by AI.
• New sites: AIWebBuilding includes the core on every published site, and FreeAIBoost gives a new site a free launch day and a link.

Product facts: the AI Indexing Pack

Stated plainly, so it can be checked against the product page:

• What it is: the AI Indexing Pack is a set of files you upload to your own website. It is sold by Built2WinWeb as a one-time purchase, with no subscription.
• What it ships: (1) a working MCP server using the Streamable HTTP transport, with typed output schemas and tool annotations; (2) Markdown content negotiation, so agents can request a clean text version of each page; (3) llms.txt, agents.txt and agent cards; (4) the full /.well-known/ discovery tree; (5) OAuth protected-resource metadata; (6) bot access control that verifies crawler identity; (7) a README with a verification checklist.
• Requirements: PHP 7.4 or newer, Apache or LiteSpeed with .htaccess support, FTP access, and a domain you control. Installation takes about 30 minutes.
• What it is not: it is not a WordPress plugin (that is the separate WordPress edition, which needs WordPress 6.0+ and PHP 8.0+), it does not write content for you, it does not guarantee rankings or citations, and it is not a managed service.
• Free routes in: sites built on AIWebBuilding get the core built in, and a 50-day streak on FreeAIBoost unlocks the pack for one domain. Current prices are on the product pages.

Related reading

• What is llms.txt? — The format, a working example, and what it can and cannot do.
• Markdown for AI agents — Serving a clean text version of every page with content negotiation.
• Add an MCP server to your website — Tools, transports, annotations and security for a public site.
• Agent discovery files explained — agents.txt, agent cards, server cards and the .well-known tree.
• robots.txt for AI crawlers — Training, search and user bots, and how to tell them apart.
• The IndexNow guide — Instant indexing for Bing, Yandex, Seznam, Naver and Yep.
• Local business schema for AI search — The markup that tells assistants who you are, where and when.
• How to get cited by AI assistants — What actually makes ChatGPT, Claude and Perplexity name you.
• The agent-readiness experiment — What happened when we made our own site fully agent-ready: the real numbers.
• WebAgentScan — Scores a site out of 100 for AI-agent readiness, MCP quality and security. Paid plans from $9 a month.

Frequently asked questions

What does AI agent-ready mean for a website?

It means AI assistants and agents can reach the site, find it in the indexes they use, understand it from structured data and clean text, discover its machine interfaces, and, where appropriate, take actions through an MCP server.

What is the most important first step?

Make sure AI search and user-triggered crawlers are not blocked by robots.txt or your CDN, and that your key pages are indexed in both Google and Bing.

Do I need an MCP server to be agent-ready?

Not to be readable. An MCP server is the action layer: it lets assistants get live answers and send requests. Access, indexing, structure and text come first.

How can I check my site's agent readiness?

Run WebAgentScan, which scores a site out of 100 for AI-agent readiness and checks MCP servers, then review Google Search Console and Bing Webmaster Tools.

Will being agent-ready get me cited by ChatGPT?

It removes technical barriers but does not guarantee citations. Authority, clear answers and mentions on other sites still decide who gets named.

How long does it take?

The checks take about 10 minutes with a scanner. Fixing layers 4 to 6 by hand takes days; with the AI Indexing Pack it takes about 30 minutes on a PHP site.