AX Audit vs. AgentGrade

AgentGrade's own site describes five scoring dimensions — Discovery, Capability exposure, Content negotiation, Payment protocols, and Trust & identity — checking protocols like MCP, x402, llms.txt, OpenAPI, WebMCP, A2A, and SKILL.md. That's a wide protocol sweep. What AgentGrade's public pages don't mention anywhere is agent-initiated account creation: whether an agent can actually sign up, get provisioned, and start using your service without a human clicking through a form. That's the gap AX Audit's AR-PROV category exists to close.

Can an agent actually open an account with you?

Every tool on this page — including this one — checks whether agents can find and read your site, and most now check whether they can transact with it. None of the three we compare against here define or score agent-initiated account provisioning: machine-readable terms, credential issuance, and signed-mandate consent, so an agent can sign up and get configured without a human filling out a form. AX Audit calls that AR-PROV. It's AINative's own extension to the taxonomy, not a repackaging of someone else's check.

Verified via AgentGrade's own site (agentgrade.com/agent-readiness and homepage): the five named dimensions, the specific protocols listed per dimension, and that pricing/payment gating is "planned but not currently active" (i.e. currently free). AgentGrade publicly claims "70+ checks" — we report that as their claim, not ours, and don't restate it as a verified AINative number. We found no mention of account provisioning anywhere on their public pages.

See where your site actually stands

Free scan, about 60 seconds. Six categories plus AR-PROV — the one none of these tools check yet.