← Back to Architecture
Architecture
Risk Engine
Decision logic for challenge strength and escalation.
Latest version
AGVS Draft v0.1This section is part of the current public OSCF documentation draft.
Inputs
- request frequency
- retry count
- challenge reuse attempts
- session freshness
- automation indicators
- device consistency signals
Outputs
- risk_score
- risk_level
- challenge_strength
- allow / retry / escalate / block
Risk levels
Low
Normal interaction pattern, no suspicious retries.
Medium
Elevated retry count or unusual timing.
High
Replay attempt detected or rapid abuse pattern.