{
  "title": "tau-rag · Overview",
  "uid": "tau-rag-overview",
  "schemaVersion": 39,
  "version": 1,
  "refresh": "30s",
  "tags": ["tau-rag", "rag"],
  "time": {"from": "now-6h", "to": "now"},
  "panels": [
    {
      "type": "stat",
      "title": "Request Rate (qps)",
      "targets": [{
        "expr": "sum(rate(tau_rag_requests_total[5m]))",
        "refId": "A"
      }],
      "gridPos": {"x": 0, "y": 0, "w": 6, "h": 4}
    },
    {
      "type": "stat",
      "title": "5xx Error Rate",
      "targets": [{
        "expr": "sum(rate(tau_rag_requests_total{status=~\"5..\"}[5m])) / sum(rate(tau_rag_requests_total[5m]))",
        "refId": "A"
      }],
      "fieldConfig": {"defaults": {"unit": "percentunit"}},
      "gridPos": {"x": 6, "y": 0, "w": 6, "h": 4}
    },
    {
      "type": "stat",
      "title": "p95 Latency (s)",
      "targets": [{
        "expr": "histogram_quantile(0.95, sum(rate(tau_rag_request_duration_seconds_bucket[5m])) by (le))",
        "refId": "A"
      }],
      "fieldConfig": {"defaults": {"unit": "s"}},
      "gridPos": {"x": 12, "y": 0, "w": 6, "h": 4}
    },
    {
      "type": "stat",
      "title": "p99 Latency (s)",
      "targets": [{
        "expr": "histogram_quantile(0.99, sum(rate(tau_rag_request_duration_seconds_bucket[5m])) by (le))",
        "refId": "A"
      }],
      "fieldConfig": {"defaults": {"unit": "s"}},
      "gridPos": {"x": 18, "y": 0, "w": 6, "h": 4}
    },
    {
      "type": "timeseries",
      "title": "Requests by endpoint",
      "targets": [{
        "expr": "sum(rate(tau_rag_requests_total[1m])) by (endpoint)",
        "refId": "A", "legendFormat": "{{endpoint}}"
      }],
      "gridPos": {"x": 0, "y": 4, "w": 24, "h": 8}
    },
    {
      "type": "timeseries",
      "title": "Retrieval latency (p50/p95/p99)",
      "targets": [
        {"expr": "histogram_quantile(0.5, sum(rate(tau_rag_retrieval_duration_seconds_bucket[5m])) by (le))", "refId": "A", "legendFormat": "p50"},
        {"expr": "histogram_quantile(0.95, sum(rate(tau_rag_retrieval_duration_seconds_bucket[5m])) by (le))", "refId": "B", "legendFormat": "p95"},
        {"expr": "histogram_quantile(0.99, sum(rate(tau_rag_retrieval_duration_seconds_bucket[5m])) by (le))", "refId": "C", "legendFormat": "p99"}
      ],
      "gridPos": {"x": 0, "y": 12, "w": 12, "h": 8}
    },
    {
      "type": "timeseries",
      "title": "Generation latency (p50/p95/p99)",
      "targets": [
        {"expr": "histogram_quantile(0.5, sum(rate(tau_rag_generation_duration_seconds_bucket[5m])) by (le))", "refId": "A", "legendFormat": "p50"},
        {"expr": "histogram_quantile(0.95, sum(rate(tau_rag_generation_duration_seconds_bucket[5m])) by (le))", "refId": "B", "legendFormat": "p95"},
        {"expr": "histogram_quantile(0.99, sum(rate(tau_rag_generation_duration_seconds_bucket[5m])) by (le))", "refId": "C", "legendFormat": "p99"}
      ],
      "gridPos": {"x": 12, "y": 12, "w": 12, "h": 8}
    },
    {
      "type": "timeseries",
      "title": "Answer quality signals",
      "targets": [
        {"expr": "tau_rag_answer_grounding_ratio", "refId": "A", "legendFormat": "grounding"},
        {"expr": "tau_rag_numeric_support_ratio", "refId": "B", "legendFormat": "numeric"},
        {"expr": "tau_rag_coverage_ratio", "refId": "C", "legendFormat": "coverage"}
      ],
      "gridPos": {"x": 0, "y": 20, "w": 12, "h": 8}
    },
    {
      "type": "stat",
      "title": "Flaky queries (v3.64)",
      "targets": [{"expr": "tau_rag_flaky_queries_count", "refId": "A"}],
      "gridPos": {"x": 12, "y": 20, "w": 6, "h": 4}
    },
    {
      "type": "stat",
      "title": "ECE (v3.55)",
      "targets": [{"expr": "tau_rag_calibration_ece", "refId": "A"}],
      "gridPos": {"x": 18, "y": 20, "w": 6, "h": 4}
    }
  ]
}