Login
Confirm the destination before credentials leave the agent.
Insert a decision checkpoint immediately before a sensitive action. Trustiry returns a stop or allow result without hiding the evidence gaps behind a single score.
The same destination can require different handling depending on what the agent is about to do.
Confirm the destination before credentials leave the agent.
Stop checkout when a reviewed relationship is missing or risk is high.
Keep identity files and business documents away from unverified destinations.
Inspect the receiver before sending personal or operational data.
Avoid establishing a durable account relationship with an impersonator.
POST https://api.trustiry.com/v1/agent-safety
Authorization: Bearer <tr_live_your_key>
Content-Type: application/json
{
"url": "https://amaz0n-secure-login.example/checkout",
"intended_action": "payment"
}{
"decision_id": "9d6f…",
"normalized_domain": "amaz0n-secure-login.example",
"relationship_status": "suspected_impersonation",
"risk_status": "high_risk_signal_observed",
"action": "payment",
"action_allowed": false,
"reason_codes": ["brand_impersonation_signal"],
"official_alternative": "https://www.amazon.com",
"confidence_class": "high",
"unknowns": [],
"expires_at": "2026-07-19T15:00:00.000Z"
}These rules prevent a useful signal from turning into a misleading permanent label.
Treat action_allowed: false as a hard stop for the requested action.
Preserve reason_codes, unknowns, confidence_class, and the dispute path in user-visible decisions.
Refresh after expires_at. Never store a time-bounded decision as a permanent verdict.
Use POST and avoid copying paths or query strings into logs, analytics, or support tools.