Scoring model
Risk scores combine reputation, infrastructure, redirect analysis, domain age, certificate signals, and graph associations.
Confidence calculation
Confidence is separate from risk. A highly confident benign result can still be safe, while an uncertain result may require more corroboration.
Intelligence weighting
Weights are adjusted by signal type and source reliability.
- Observed infrastructure carries more weight than inferred context.
- Strong provider corroboration raises confidence.
- Conflicting evidence lowers certainty, not just score.
Sample table
| Signal | Weight | Impact |
|---|---|---|
| Newly registered domain | 0.25 | Raises suspicion |
| Malicious redirect chain | 0.40 | Strong risk indicator |
| Known bad ASN | 0.20 | Infrastructure context |
| Certificate mismatch | 0.15 | Trust degradation |
AI-assisted analysis
The model does not replace analyst judgment. It helps explain the signal mix, summarize evidence, and surface likely relationships to review.
Analyst-friendly output
Hackura Sentinel AI should feel like a system that helps users reason about risk instead of hiding the logic behind a black box.