Trustli · Agent Verification Protocol v0.1
The same 15 checks an independent verification runs — answered by you, about your agent. Fifteen questions, about four minutes. Nothing is sent anywhere; this runs entirely in your browser.
Free · No signup · Nothing leaves this pageCan the agent operate safely when things go wrong, and is there a record? Any failure here caps the verdict — an agent with no guardrails or no log cannot be verified, whatever else it does well.
When nobody is watching, does something other than luck decide the agent is finished — and is "done" proven by evidence the agent can't rephrase?
Can what the agent claims it did be checked against a record it doesn't write itself? The last three apply only to multi-agent systems.
Answer the checks above to see where you stand.
Which is exactly what your customer's security team can't accept — every vendor says the same thing. Verification is an independent party checking the same 15 points against real evidence: a staging copy of your agent, live challenge testing by probe agents built on different models, and a deterministic verdict signed by a named human. Your buyers get a public badge page they can audit in minutes.
The first scan Trustli ever ran was on my own agents. They failed 6 of the 15 checks, and the results are public. Independent means the verdict can hurt. — Paul Hopcraft
What this tool is, precisely. A self-assessment. It scores your own answers against the published protocol using fixed rules — no AI judges anything here, and no answer is stored, transmitted, or seen by anyone but you. It cannot verify a single claim you make, and it does not produce a badge or a verdict. That gap between what you assert and what an independent party confirms is the entire reason verification exists.