Cloudflare Agent Readiness Score Review: Does It Measure AI Visibility?

What does it check?

Cloudflare's launch post lists four scored dimensions, plus commerce checks that do not count towards the score:

llms.txt is not checked by default, though you can add it to a scan. Results are also available in the Agent Readiness tab of URL Scanner, through the URL Scanner API, and from an MCP server with a scan_site tool, so an agent can run scans itself.

What did Cloudflare find across the web?

Cloudflare Radar scanned the 200,000 most visited domains, after removing categories such as redirects and ad servers, and published an adoption chart that updates weekly. At launch:

• 78% of sites had a robots.txt, but most were written for search crawlers, not AI agents.
• 4% declared AI usage preferences with Content Signals.
• 3.9% passed Markdown content negotiation.
• MCP Server Cards and API Catalogs appeared on fewer than 15 sites in the whole dataset.

Low scores are therefore normal. A site that fails most capability checks is in the same position as nearly every other site.

What does it do well?

• Free and quick: enter a URL and get a report in the browser.
• Actionable: each failing check comes with a prompt for a coding agent to implement the fix.
• Scriptable: the MCP server and URL Scanner API let agencies and developers scan many domains.
• Benchmarked: the Radar chart shows how adoption of each standard changes over time.
• Educational: the launch post and linked skill documents explain each standard clearly.

Where are its limits?

• It measures standards, not outcomes. No check asks an AI assistant anything about your brand, so a perfect score says nothing about whether you are recommended.
• Several standards are early. The MCP Server Card is a draft proposal, the Agent Skills index is Cloudflare's own proposal, and WebMCP is in an early origin trial. Adoption is tiny.
• Not every check applies to every site. Cloudflare itself says Web Bot Auth only matters if your site runs agents that make requests to other sites. API catalogs and OAuth discovery matter only if you offer APIs or sign-in for agents.
• Agent support is uneven. A February 2026 test by Checkly, cited in Cloudflare's launch post, found that only three of seven coding agents (Claude Code, OpenCode and Cursor) requested Markdown by default.
• It leans towards Cloudflare's own features. Markdown for Agents, Content Signals, Web Bot Auth and x402 are all standards Cloudflare builds or promotes. They are open, but the list reflects Cloudflare's view of what matters.

Does a higher score improve AI visibility?

There is no evidence that it does on its own. In controlled tests for Vercel's August 2026 guide with ora.ai (190 fetch probes across 19 site setups), adding llms.txt, JSON-LD, sitemaps or Markdown mirrors did not change whether agents could retrieve an answer already present in server-rendered HTML. AI visibility also depends on things this score does not test, such as whether crawlers can reach your pages and what other sites say about you.

Some checks do matter indirectly. AI bot rules in robots.txt decide whether AI search crawlers may read your pages, and a site that blocks them by accident is much less likely to be cited or to have its own wording used. Markdown negotiation can make pages cheaper for agents to read, as covered in how to serve Markdown to AI agents while serving HTML to browsers. Crawler controls are covered in what Cloudflare's AI Crawl Control and Pay per Crawl do.

How does it compare with other readiness checkers?

None of the readiness checkers replaces a visibility tracker. Cloudflare draws the same line itself: its AEO Suite, announced in August 2026, places Agent Readiness next to a separate AEO Visibility Dashboard, in early access, that asks Claude and GPT category questions and reports citation rate, mention rate and share of voice. The readiness score is the technical half; the dashboard is the visibility half. How trackers differ is covered in Profound vs Ahrefs for measuring AI visibility.

How should you read your result?

1. Sort the checks into those that apply to your site and those that do not, such as Web Bot Auth or API catalogs for a site with no API.
2. Fix bot access first. Confirm your robots.txt allows the AI search crawlers you want, since blocking them makes citation much less likely.
3. Then fix content format, such as Markdown negotiation, if agents read your pages often.
4. Rescan after each change, and record the date, since standards and checks will keep changing.

Who should use it?

• Developer platforms and SaaS products that want agents to use their APIs: the capability checks are a good roadmap.
• Publishers deciding how agents may use content: the Content Signals and bot rules checks are directly useful.
• Agencies auditing technical access at scale, using the API or MCP server.
• Most small business sites only need the discoverability and bot access checks. Ignore the rest until your platform adds them. When an MCP server is worth building is covered in whether websites need MCP, and the files involved in what goes in a site's .well-known directory.

We have not compared Agent Readiness scores with AI citation rates across sites. This review is based on Cloudflare's launch post and our reading of the checks it describes.

What might this look like in practice?

Consider an illustrative accounting software company, Plovby, whose score came back low. The report flagged robots.txt with no AI bot rules, no Markdown negotiation, and no API catalog, OAuth discovery or MCP Server Card. The team added explicit rules allowing AI search crawlers, turned on Markdown responses at its CDN, and published an API catalog because it already had a public API. It left the MCP Server Card until its MCP server shipped. The score rose, but the marketing team kept measuring AI mentions separately, since the score could not tell them whether assistants recommended Plovby.

What do people ask about the Agent Readiness score?

Is Cloudflare's Agent Readiness score free?

Yes. You enter a URL at isitagentready.com, and the same checks are in Cloudflare's URL Scanner.

Does it check llms.txt?

Not by default. You can add llms.txt to a custom scan.

Do commerce protocols affect the score?

No. It checks x402, UCP and ACP support but does not count them towards the score.

Will a perfect score get me cited by ChatGPT?

No. The score measures technical standards. It does not test whether crawlers can reach your pages or what other sites say about you, and no score can guarantee a citation.