# ─────────────────────────────────────────────────────────────────────────────
# Coralogix Custom Dashboard – Agent Monitor
#
# Import via Coralogix UI: Dashboards → Custom Dashboards → Import
# Or via API:
#   curl -X POST "https://api.coralogix.com/api/v1/external/grafana/api/dashboards/db" \
#     -H "Authorization: Bearer $CORALOGIX_API_KEY" \
#     -H "Content-Type: application/json" \
#     -d @dashboards.yaml
#
# This dashboard mirrors the Grafana dashboard (../grafana/dashboards/) while
# leveraging Coralogix-native features: DataPrime queries, log correlation,
# distributed tracing waterfall, and Apdex scoring.
# ─────────────────────────────────────────────────────────────────────────────

dashboard:
  name: "Agent Monitor – Operations"
  description: "Claude Code Agent Monitor: real-time operations, SLOs, and infrastructure health"
  folder: "Agent Monitor"
  tags:
    - agent-monitor
    - operations
    - sre

  # ── Row 1: Overview ──────────────────────────────────────────────────────
  rows:
    - name: "Overview"
      panels:
        - title: "Active Sessions"
          type: line-chart
          query:
            type: metrics
            promql: 'agent_monitor_active_sessions'
            legend: "{{namespace}}"
          span: 4

        - title: "Request Rate (req/s)"
          type: line-chart
          query:
            type: metrics
            promql: 'sum(rate(http_requests_total{job="agent-monitor"}[5m]))'
            legend: "Requests/sec"
          span: 4

        - title: "WebSocket Connections"
          type: line-chart
          query:
            type: metrics
            promql: 'websocket_connections_active{job="agent-monitor"}'
            legend: "{{pod}}"
          span: 4

    # ── Row 2: HTTP Performance ────────────────────────────────────────────
    - name: "HTTP Performance"
      panels:
        - title: "Latency Distribution (P50 / P95 / P99)"
          type: line-chart
          query:
            type: metrics
            promql: |
              histogram_quantile(0.50, sum(rate(http_request_duration_seconds_bucket{job="agent-monitor"}[5m])) by (le))
              histogram_quantile(0.95, sum(rate(http_request_duration_seconds_bucket{job="agent-monitor"}[5m])) by (le))
              histogram_quantile(0.99, sum(rate(http_request_duration_seconds_bucket{job="agent-monitor"}[5m])) by (le))
          span: 6

        - title: "Error Rate (%)"
          type: line-chart
          query:
            type: metrics
            promql: |
              sum(rate(http_requests_total{job="agent-monitor", status=~"5.."}[5m]))
              /
              sum(rate(http_requests_total{job="agent-monitor"}[5m])) * 100
            legend: "5xx %"
          thresholds:
            - value: 1
              color: yellow
            - value: 5
              color: red
          span: 3

        - title: "Status Code Distribution"
          type: bar-chart
          query:
            type: metrics
            promql: 'sum by (status) (increase(http_requests_total{job="agent-monitor"}[1h]))'
          span: 3

    # ── Row 3: Logs (DataPrime) ────────────────────────────────────────────
    - name: "Application Logs"
      panels:
        - title: "Error Logs"
          type: dataprime
          query: |
            source logs
            | filter $d.cx.application.name == 'agent-monitor'
            | filter $d.severity == 'ERROR' || $d.severity == 'FATAL'
            | select $m.timestamp, $d.k8s.pod.name, $d.message
            | order by $m.timestamp desc
            | limit 100
          span: 6

        - title: "Log Volume by Severity"
          type: bar-chart
          query:
            type: dataprime
            expression: |
              source logs
              | filter $d.cx.application.name == 'agent-monitor'
              | count_group_by $d.severity as count
          span: 3

        - title: "Hook Event Throughput"
          type: line-chart
          query:
            type: dataprime
            expression: |
              source logs
              | filter $d.cx.application.name == 'agent-monitor'
              | filter $d.message matches 'hook.*event'
              | count_per_time 1m as throughput
          span: 3

    # ── Row 4: Infrastructure ──────────────────────────────────────────────
    - name: "Infrastructure"
      panels:
        - title: "CPU Usage (%)"
          type: line-chart
          query:
            type: metrics
            promql: |
              rate(container_cpu_usage_seconds_total{
                namespace=~"agent-monitor.*",
                container="agent-monitor"
              }[5m]) * 100
            legend: "{{pod}}"
          span: 4

        - title: "Memory Usage (MiB)"
          type: line-chart
          query:
            type: metrics
            promql: |
              container_memory_working_set_bytes{
                namespace=~"agent-monitor.*",
                container="agent-monitor"
              } / 1024 / 1024
            legend: "{{pod}}"
          span: 4

        - title: "Pod Status"
          type: gauge
          query:
            type: metrics
            promql: |
              count by (phase) (
                kube_pod_status_phase{namespace=~"agent-monitor.*"}
              )
          span: 4

    # ── Row 5: Database & Storage ──────────────────────────────────────────
    - name: "Database & Storage"
      panels:
        - title: "SQLite Query Duration (ms)"
          type: line-chart
          query:
            type: metrics
            promql: 'sqlite_query_duration_seconds{job="agent-monitor"} * 1000'
          span: 4

        - title: "PV Usage (%)"
          type: gauge
          query:
            type: metrics
            promql: |
              (kubelet_volume_stats_used_bytes{namespace=~"agent-monitor.*"}
              / kubelet_volume_stats_capacity_bytes{namespace=~"agent-monitor.*"}) * 100
          thresholds:
            - value: 70
              color: yellow
            - value: 90
              color: red
          span: 4

        - title: "Network I/O (bytes/s)"
          type: line-chart
          query:
            type: metrics
            promql: |
              sum by (pod) (rate(container_network_receive_bytes_total{namespace=~"agent-monitor.*"}[5m]))
              sum by (pod) (rate(container_network_transmit_bytes_total{namespace=~"agent-monitor.*"}[5m]))
          span: 4

    # ── Row 6: SLO Tracking ────────────────────────────────────────────────
    - name: "SLO Tracking"
      panels:
        - title: "Availability SLO (99.9% target)"
          type: gauge
          query:
            type: metrics
            promql: |
              (1 - sum(rate(http_requests_total{job="agent-monitor", status=~"5.."}[30d]))
              / sum(rate(http_requests_total{job="agent-monitor"}[30d]))) * 100
          thresholds:
            - value: 99.9
              color: green
            - value: 99.5
              color: yellow
            - value: 99.0
              color: red
          span: 4

        - title: "Latency SLO (P95 < 500ms)"
          type: gauge
          query:
            type: metrics
            promql: |
              histogram_quantile(0.95,
                sum(rate(http_request_duration_seconds_bucket{job="agent-monitor"}[30d])) by (le)
              ) * 1000
          thresholds:
            - value: 300
              color: green
            - value: 500
              color: yellow
            - value: 1000
              color: red
          span: 4

        - title: "Error Budget Remaining"
          type: gauge
          query:
            type: metrics
            promql: |
              (0.001 - sum(rate(http_requests_total{job="agent-monitor", status=~"5.."}[30d]))
              / sum(rate(http_requests_total{job="agent-monitor"}[30d])))
              / 0.001 * 100
          thresholds:
            - value: 50
              color: green
            - value: 25
              color: yellow
            - value: 0
              color: red
          span: 4