Agent Continuous
Trust Monitoring
An evaluation score at deployment is a point-in-time measurement. As your agent operates in production, its behavior and trust posture change. NGIQ Continuous Monitoring tracks that change — and acts before it becomes an incident.
Required for L4 (High Assurance)
Agents seeking NGIQ Trust Level L4 must enroll in continuous monitoring as part of the KYA Verified evaluation. L4 cannot be issued or maintained without an active monitoring integration. L2 and L3 agents may enroll voluntarily to strengthen their published trust profile.
What Monitoring Covers
Six integrated capabilities — from heartbeat to credential lifecycle.
Heartbeat Integration
Your agent posts to POST /api/monitor/heartbeat at a configured interval. Missed heartbeats trigger immediate alert escalation. No polling on your side — we handle the monitoring infrastructure.
Composite Score Tracking
Every 5 minutes, your agent's operational score is recalculated from uptime, transaction success rate, dispute rate, and response consistency. Published score updates combine evaluation score and operational score.
Anomaly Detection
Gemini-powered risk classification runs on every score event. Anomalies are classified LOW / MEDIUM / HIGH. High-risk patterns trigger automatic protective action without waiting for human review.
Dispute Filing
Counterparties, users, or operators can file disputes against an agent's behavior via POST /api/monitor/dispute. Disputes feed directly into the composite score and trigger review workflows.
Transaction Outcome Reporting
Agents report transaction outcomes via POST /api/monitor/outcome. Success rates, failure modes, and latency patterns are analyzed continuously to detect behavioral drift from declared capabilities.
Credential Lifecycle Management
Monitoring integrates with the NGIQ verifiable credential. Score drift beyond thresholds triggers automatic credential status updates — suspension for MEDIUM-HIGH anomalies, revocation for CRITICAL events.
Tiered Drift Response
Not all anomalies require the same response. The monitoring system applies graduated responses based on severity.
Log event. No alert. Included in weekly digest.
Email alert to operator + Madhu (NextGenIQ). Agent flagged for review within 48 hours.
Immediate email + Slack. Agent credential suspended automatically. Operator must remediate before reinstatement.
Credential revoked automatically. Public verification endpoint updated. Operator notified immediately with incident report.
Published Score Methodology
An agent's published NGIQ Trust Level reflects both its evaluation score and its real-world operational track record. Evaluation score and operational performance are combined into a single published score — the exact weighting formula is proprietary and not exposed via API.
A published trust level can only decrease without a new evaluation. Score drift of more than 5 points triggers a warning; drift beyond 15 points triggers a critical alert and credential review.
Enterprise Sandbox Onboarding
Enterprise teams deploying multiple agents can enroll in the NextGenIQ Monitoring Sandbox — a dedicated environment to configure integrations, test heartbeat and outcome reporting, and validate drift detection before going live. Sandbox access is included with KYA Verified evaluations.
- Dedicated test endpoints for heartbeat and outcome reporting
- Simulated drift scenarios to validate alert configuration
- Monitoring dashboard preview with sample data
- Integration review call with NextGenIQ team
Enroll Your Agent in Monitoring
Monitoring enrollment begins with a KYA Verified evaluation. Existing KYA Verified clients can add monitoring at any time.