Risk Scoring

Arrakis computes a 0-100 risk score per agent based on active security findings across multiple weighted security domains. The scoring model emphasizes worst-case findings: a single critical vulnerability drives the score regardless of how healthy other areas are. Scores update in real time as alerts are raised or resolved.

Agents are classified into risk levels based on their score: Critical, High, Medium, Low, and None. These levels drive dashboard views, alert prioritization, and policy enforcement decisions.

The core principle is that security risk is not well described by averages. An agent with perfect scores in four domains and a critical vulnerability in one is not “mostly safe.” The scoring model reflects this.

Full scoring documentation is available to Arrakis customers.