1 files changed, 50 insertions, 0 deletions
diff --git a/docs/reference/grafana-dashboard.md b/docs/reference/grafana-dashboard.md
new file mode 100644
index 0000000..b7f2030
--- /dev/null
+++ b/docs/reference/grafana-dashboard.md
@@ -0,0 +1,50 @@
+# Grafana Dashboard
+
+A dashboard is provided that shows all Epimetheus test metrics.
+
+## Panels
+
+1. Request Rate (line graph)
+2. Total Requests (stat)
+3. Active Connections (gauge with thresholds)
+4. Temperature (gauge with thresholds)
+5. Request Duration Histogram (p50, p90, p99)
+6. Average Request Duration (stat)
+7. Jobs Processed by Type (bar gauge)
+8. Jobs Status Breakdown (table)
+
+Auto-refresh: 10 seconds. Time range: last 15 minutes (configurable). Optimized for dark theme.
+
+## Deployment
+
+### Option 1: Kubernetes ConfigMap (recommended)
+
+If you have a manifest that defines the dashboard as a ConfigMap with Grafana’s discovery label:
+
+```bash
+kubectl apply -f ../prometheus/epimetheus-dashboard.yaml
+```
+
+Grafana will pick it up automatically.
+
+### Option 2: Manual import
+
+1. Port-forward Grafana: `kubectl port-forward -n monitoring svc/prometheus-grafana 3000:80`
+2. Open http://localhost:3000
+3. Dashboards → Import → Upload `grafana-dashboard.json`
+
+### Option 3: Deploy script
+
+```bash
+./scripts/deploy-dashboard.sh
+# Or with credentials:
+GRAFANA_URL="http://localhost:3000" GRAFANA_USER="admin" GRAFANA_PASSWORD="yourpassword" ./scripts/deploy-dashboard.sh
+```
+
+## Datasource
+
+Use Prometheus (or a Prometheus-compatible backend such as VictoriaMetrics) as the datasource. Point it at the same instance Epimetheus writes to (e.g. http://localhost:9090 after port-forward).
+
+## Panel guidelines
+
+When creating or updating Grafana panels, follow the project’s [AGENT.md](../../AGENT.md) (Grafana dashboard guidelines): e.g. sort time series by last value descending, use `sort_desc()` in bar gauges, set table sort options as specified.