Confidence Scores
Understanding how HealBot evaluates anomalies mapping strict safety execution boundaries evaluating explicit scoring metrics.
How Scores Are Calculated
HealBot analyzes error footprints identifying anomaly bounds checking historical patterns, correlating multi-service signals, and extracting clear component isolations scaling precision. A score out of 100% strictly determines resolution confidence.
Score Thresholds
Clear problem isolation mapping robust historical successes executing perfectly isolated patches natively.
Similar resolutions successfully executed mapping slight variance across connected instances.
Unresolved architecture bugs lacking historic contexts demanding manual review approval thresholds.
Low confidence mapping unknown pattern exceptions skipping unapproved automated executions completely.
Autonomy Overrides
CartNerve settings explicitly allow redefining Autonomy Thresholds. If a generated confidence score meets or exceeds the defined Autonomy Threshold, HealBot resolves the problem mapping Auto-Heal functions immediately avoiding human interference matrices. Otherwise, it sits in Pending Approval.