Hide and Seek vs. Cloudflare
Cloudflare’s Is Your Site Agent-Ready? and Hide and Seek both scan a site and give it a score out of 100. They answer different questions, so the same site can score low on one and high on the other, and both numbers can be right.
Two questions
| Is Your Site Agent-Ready? | Hide and Seek | |
|---|---|---|
| Asks | Can an AI agent use your site? | Can AI find you, and does it? |
| For | Sites with an API, a login, or a store that agents act on | Anyone who wants to be the answer when someone asks AI |
| Reads | robots.txt, response headers, DNS, and well-known files for new agent standards | Your pages: whether AI can reach them, read them, tell who you are, and find all of them |
| Who you are | Not checked | Title and description, structured identity, profiles that link back, an about page, Wikidata, and directory listings |
| Whether AI names you | Not checked | Found: every month, ChatGPT, Claude, Gemini, and Perplexity are asked what you want to be found for |
| Over time | A scan when you ask | A free report stands for a week; plans rescan every week and keep the record |
| Fixes | A prompt for your coding agent with each check | The fixes worth the most first, and one prompt for the whole report |
What Cloudflare checks
Mostly the plumbing of agent standards that are still being written: Link headers, DNS records for agent discovery, an API catalog, OAuth discovery, an MCP server card, an agent skills index, WebMCP, and protocols for agents that buy things. If agents will log in to your site, call your API, or check out from your store, those matter, and Cloudflare’s scan is the right place to start.
If you’re a studio, a writer, a shop, or a product without a public API, most of them don’t apply to you, and missing them won’t keep AI from naming you.
What the Hide and Seek Rubric asks
The rubric’s criteria ask whether AI can reach your pages, read them, tell who you are, and find all of them. Where the two scans overlap (robots.txt, a sitemap, whether AI’s crawlers are allowed in, Markdown for agents, and Content Signals), the rubric asks too; the last two are reported, not scored.
Then the part no scan can do. A scan says whether AI can find you. Found says whether it does: who gets named when someone asks the question you want to be the answer to, and whether you’re among them.
Use both
If agents will act on your site, run Cloudflare’s scan. If you want to be the answer, run ours. If you’re both, run both. The two scores won’t agree, and they shouldn’t.