AI Product Index / How agent-readiness checkers differ
There are several ways to be told whether AI agents can read your site, and they disagree — not about the facts, but about what counts. This page is the honest map, including where this one is the weaker choice.
| checker | what it is | how you use it |
|---|---|---|
| Cloudflare Agent Readiness | The most complete published checklist: discoverability, content, bot access control, capabilities, and commerce as a checked-but-unscored dimension. | Tied to Cloudflare's own tooling and audience. The checklist itself is public and worth reading whoever hosts you. |
| agentswelcome.dev | 18 checks and a certification gate at 70/100, plus a public directory of sites that pass. | Pure API, no account: POST a URL and it audits you live. The directory listing is the point — it is a place to be found. |
| This site | 20 weighted checks across two published generations, with every weight and remedy documented at /checks/. | Free grade over GET, paid per-check fixes over HTTP 402. Callable by an agent without a human present, which is the part the others do not do. |
Run all three. They are free, they take a minute each, and they overlap enough that anything all three flag is real. Where they differ is mostly recency: checklists have been moving fast, several specs changed paths in the last year, and a checker that has not been updated recently will pass you on a path a current client no longer reads. That is the failure mode worth checking for in any of them, including this one — which is why the check set here is versioned in public.
Try the free grade · read the checklist · see how others score.