Threat Detection & Response Procedures#
High-Severity Anomaly Response#
Detection Criteria#
- Security event spike > 3-sigma deviation from baseline
- Critical severity events > threshold
- Multiple simultaneous threat types
Response Procedure#
-
Immediate Assessment (0-5 minutes)
- Access security dashboard:
GET /v1/security/dashboard - Identify affected services and threat types
- Check system health:
GET /v1/health
- Access security dashboard:
-
Threat Analysis (5-15 minutes)
- Correlate with application logs
- Check rate limiter telemetry
- Analyze traffic patterns in WAF/CDN
-
Containment (15-30 minutes)
- Apply rate limiting if needed
- Block malicious IPs via WAF
- Scale affected services if under load
-
Investigation (30-60 minutes)
- Deep dive into affected user accounts
- Check for credential stuffing patterns
- Analyze attack vectors and entry points
Escalation Criteria#
- Critical severity events > 10/hour
- Multiple service compromise indicators
- Data exfiltration attempts detected
Authentication Failure Investigation#
Common Patterns#
- Credential Stuffing: High volume auth failures from multiple IPs
- Brute Force: Repeated failures from single IP
- Account Takeover: Successful auth after multiple failures
Investigation Steps#
- Query auth service logs:
kubectl logs -n lilith deployment/auth-service - Check rate limiting metrics
- Analyze IP geolocation patterns
- Review affected user accounts
Mitigation Actions#
- Temporarily block suspicious IPs
- Force password reset for affected accounts
- Enable additional MFA requirements
- Update rate limiting thresholds
Safety Policy Violations#
Detection#
safety.policy_violationeventsmoderation.flaggedcontent- Persona boundary violations
Response Process#
- Review flagged content in moderation queue
- Analyze user interaction patterns
- Update safety filters if needed
- Escalate to safety team for review
Automated Response Triggers#
High-Priority Alerts#
- Critical security events: Immediate PagerDuty alert
- Anomaly detection: Slack notification
- Service health degradation: Auto-scaling trigger
Response Automation#
javascript
// Example automated response configuration
const responseConfig = {
'rate_limit_exceeded': {
severity: 'medium',
action: 'increase_rate_limits',
cooldown: '5m'
},
'jailbreak_attempt': {
severity: 'high',
action: 'alert_security_team',
escalation: 'immediate'
},
'auth_failed': {
severity: 'low',
action: 'track_ip_patterns',
threshold: 50
}
};
Monitoring Commands#
Dashboard Access#
bash
# Security dashboard
curl ${OBSERVABILITY_URL}/v1/security/dashboard
# Raw security events
curl ${OBSERVABILITY_URL}/v1/security/events?timeframe=1h
# Anomaly details
curl ${OBSERVABILITY_URL}/v1/security/anomalies
Kubernetes Investigation#
bash
# Check security-related pods
kubectl get pods -n lilith -l security=monitoring
# View security event logs
kubectl logs -n monitoring deployment/observability-service | grep security
# Check network policies
kubectl get networkpolicies -n lilith
Generated automatically by security-docs-generator.js Last updated: 2025-09-27T18:05:41.712Z