🎚️ Adaptive risk-based auth
Real IAM doesn't treat every request the same. This engine combines signals the Lab already produces elsewhere — device recognition, ITDR alert history, sign-in velocity, and step-up freshness — into ONE explainable score, and turns that score into a decision: allow step up block.
Honest simulation — read this first
Four of the five signals below are 100% real, derived live from state the Lab actually stores:
whether your lab_device cookie matches a recognized device, your
own unacknowledged security alerts, how many risk-relevant calls you've made in the last minute,
and whether your session holds a fresh TOTP step-up. The fifth — impossible travel — has no real IP-geolocation behind it anywhere in this Lab; it's a toggle standing in for "the
geo-IP signal flipped since your last sign-in", the exact shape a real fraud engine feeds in from
a geo-IP lookup, with the lookup itself simulated so the lesson needs no network dependency.
1. Your live risk assessment
Read-only — recomputed from your real state on every refresh.
2. Try the protected action
A demo sensitive action — nothing real happens, but it is gated by the SAME decision above. The toggle in step 1 carries through: retry it to see allow → step up → block.