Glossary

These definitions are canonical across MaiGuard documentation. For implementation details, follow the linked guides.

Shadow mode

A production-safe testing mode where new rules or models are evaluated in parallel with live rules without changing the actual ALLOW/REVIEW/BLOCK decision returned to your platform. Shadow results appear in the API response under shadowEvaluation. Shadow mode guide →

Entity graph

A network model linking customers, devices, IP addresses, counterparties, and accounts. MaiGuard uses graph traversal to detect mule networks, shared identity signals, and synthetic identity rings. Entity graph guide →

Velocity rule

A rule that triggers when transaction or login activity exceeds a threshold within a time window — for example, more than 5 transfers in 10 minutes from the same device. AML rules guide →

Behavioral biometrics

Passive signals collected from user interaction patterns — keystroke dynamics, mouse movement, scroll behavior, and session timing — used to distinguish humans from bots and detect account takeover. Behavioral biometrics guide →

PEP screening

Politically Exposed Person screening checks counterparties and beneficiaries against PEP lists as part of AML compliance workflows. MaiGuard supports tenant-managed lists and screening at scoring time. AML rules guide →

AML (Anti-Money Laundering)

Regulatory and operational controls to detect, monitor, and report suspicious financial activity. MaiGuard provides velocity rules, list screening, case management, and audit trails for AML workflows. AML rules guide →

Device fingerprinting

Identification of a browser or device using stable attributes — OS, browser version, canvas hash, fonts, and hardware signals. Used to detect returning devices and pre-transaction anomalies. JavaScript SDK →

riskScore

A numeric score from 0 to 100 returned with every scoring response. Higher values indicate greater fraud or AML risk. Rules and ML models contribute to the final score. How scoring works →

deviceSessionId

A session identifier generated by the MaiGuard JavaScript SDK when behavioral collection starts. Pass it in score requests to link browser signals with server-side scoring. Must start with ds_. SDK reference →

ALLOW / REVIEW / BLOCK

The three decision states returned by MaiGuard scoring. ALLOW means proceed; REVIEW means elevated risk requiring manual review; BLOCK means decline the transaction. How scoring works →