Ten yes/no questions about how your AI agents authenticate today — an instant score, a grade, and a fix list built from your No answers. Nothing is sent anywhere: the whole check runs in your browser, and your answers are never stored.
Every Yes is worth 10 points. A (90+) — agent-ready: own identities, scoped and expiring credentials, approval gates, full audit trail. B (75–89) — a solid foundation with a few named gaps. C (60–74) — workable but exposed. D (40–59) — at risk: too much power, held too long, with too little visibility. F (below 40) — critical: your agents are effectively logging in as you, with no safety net. The score is a checklist, not a certification — but every No on your fix list is a concrete, fixable thing.
Agent access readiness is how safely your AI agents are set up to log in and act on your behalf: whether each agent has its own scoped, expiring credentials, whether sensitive actions need your approval, and whether you can audit and revoke an agent's access quickly. This check scores those practices across ten questions.
An agent using your personal password is indistinguishable from you — to the service, to the audit log, and to an attacker. If that password leaks through a prompt, log or breach, everything on the account is exposed, and you cannot revoke the agent's access without locking yourself out and changing the password everywhere.
90 or above (grade A) means your agents have their own identities, scoped and expiring credentials, approval gates and an audit trail. 75–89 (B) is a solid foundation with a few gaps. Below 60, one leaked credential or one rogue run could do real damage — work through the fix list starting at the top.
No. The check runs entirely in your browser. Your answers are never sent to a server, stored in cookies or local storage, or tracked — reload the page and they are gone.