Projects & Services
Monitoring & Metrics
Your agent reads CPU, memory, and network metrics through MCP — observe and debug production applications programmatically
Ink collects performance metrics for every running service. Your agent reads these metrics through MCP tools to monitor health, diagnose issues, and decide when to scale.
Available metrics
| Metric | Description |
|---|---|
| CPU usage | vCPU utilization as a percentage of allocated capacity |
| Memory usage | RAM usage in MB against allocated limit |
| Network receive | Inbound traffic (bytes/sec) |
| Network transmit | Outbound traffic (bytes/sec) |
Time ranges
Metrics are available over four time periods:
| Range | Use case |
|---|---|
| 1 hour | Real-time monitoring after deployment |
| 6 hours | Short-term debugging and traffic patterns |
| 7 days | Weekly usage patterns |
| 30 days | Capacity planning and cost analysis |
Agent-driven monitoring
Your agent can check metrics via the get_service MCP tool and take action:
Agent: "Check if the API service is healthy"
→ Calls get_service(name="api", runtime_log_lines=50)
→ Sees CPU at 95%, memory at 1800/2048 MB
→ Calls update_service(name="api", memory="4096Mi", vcpus="4")
→ Service scales up automaticallyViewing in dashboard
Navigate to a service in your Ink dashboard to see performance metrics with interactive charts. The dashboard shows configured limits alongside actual usage.