How Guard works
Guard runs three processing loops at different intervals:| Loop | Interval | Purpose |
|---|---|---|
| Fast | Every 10 seconds | Critical health checks, service availability |
| Heartbeat | Every 60 seconds | Node metrics, resource usage trends |
| Deep scan | Every 6 hours | Security audit, configuration drift, updates |
Enabling Guard
Configure trust level
Choose Guard’s initial autonomy level:
- Observer — Monitor and log only (recommended to start)
- Advisor — Suggest fixes, require your confirmation
- Autonomous — Execute fixes automatically
Select monitoring targets
Choose which nodes and services Guard should monitor. You can start with a subset and expand later.
Trust progression
Guard uses a graduated trust model. Start with Observer to build confidence:What Guard monitors
- Service availability and response times
- Node resource usage (CPU, RAM, disk)
- Security events and suspicious activity
- Configuration drift from desired state
- Available system and package updates
Further reading
Trust model explained
Deep dive into the Observer-Advisor-Autonomous model
Architecture
How Guard fits into the kombify AI architecture
