Response Decoding
When ABIS speaks, it speaks in JSON. Every field carries intelligence. Every nested object reveals patterns. Master the schema, and you master the system.
ABIS responses follow a consistent JSON schema designed for immediate actionability. The root level contains the risk assessment (risk_score, risk_level, recommendation), operational metadata (request_id, timestamp, processing_time_ms), and model confidence metrics.
Nested within are behavioral_indicators arrays detailing specific patterns detected, anomaly_details objects explaining deviations from baseline behavior, and context_enrichment data showing how external factors influenced the analysis.
Responses also include next_actions arrays suggesting appropriate system responses, and debugging_info for development environments. Understanding field data types, nullable properties, and conditional presence of certain objects is critical for robust integration.
risk_assessment
Core scoring and recommendation logic. Contains score (0-100), categorical level, and suggested action.
behavioral_indicators
Array of detected patterns. Each indicator has type, severity, weight, and human-readable description.
anomaly_details
Deviation analysis from user baseline. Includes statistical significance and percentage variance.
context_enrichment
Geographic, device, temporal, and historical context used in the risk calculation.
next_actions
Recommended system responses with priority levels and implementation guidance.
metadata
Operational info: request ID, processing time, API version, and debug traces.
risk_score and anomaly_score in an ABIS response?