How can I help you today?
Ask about an incident, investigate an alert, or find an operational runbook.
Incidents —
Select an incident
Choose an incident from the list. Its details and investigation actions will appear here.
Incident
Recovery approvals
NEXUS Observability
Infrastructure, integrations and operational signals in one view.
AI Agent Performance
Live structured telemetry · rolling 24-hour window
Model activity
Execution activity
Metrics are sourced from structured CloudWatch agent logs. Percentages are calculated from the observed event samples returned by the backend.
Integration Health
Live connectivity checks
Operational Snapshot
Current data-source coverage
Agent Activity
Aggregated activity from structured agent telemetry. This view reports counts and quality signals, not a per-run trace.
Enterprise Readiness
Assess product controls, operational readiness and supporting evidence across eight enterprise domains.
Select a domain
Choose a domain above to review its controls and evidence.
Backend assessment
Live controls and evidence stored in DynamoDB. A smoke assessment checks backend table availability only.
Submit evidence
Evidence is saved as PENDING review. Submission alone does not verify a control.
Evidence register
Evidence submitted to the backend.
Assessment history
Recorded backend smoke assessments.
Windows EC2 Metrics
Live Windows EC2 infrastructure metrics retrieved through the NEXUS backend and Amazon CloudWatch.
Disk utilization
Selected Windows services
Only services configured by the NEXUS backend are shown. A stopped service is not automatically a failure; validate its startup type and operational role.
CPU utilization — Last 60 minutes
CloudWatch AWS/EC2 CPU utilization
Network traffic — Last 60 minutes
Incoming and outgoing network traffic
Week 7 — NEXUS Security Lab
Controlled red-team demonstration using synthetic inputs and local mock controls.
—
—
Not run
01 — Direct prompt injection
Attempt to override instructions and force an unsafe action.
02 — Malicious retrieved runbook
Check that malicious instructions inside retrieved text are treated as untrusted.
03 — Sensitive-data extraction
Use a synthetic secret to demonstrate response redaction.
04 — Approval bypass
Attempt a mock recovery action without a valid approval record.
Evidence log
Synthetic input, expected safeguard, observed result, and test scope.
No tests run.