Explorer mode

Explorer is the ad-hoc plotting view. Pick a metric, scope it with label filters, plot it. Use it when you want to see one specific thing over time.

Layout

Time picker at top: presets (15m | 1h | 6h | 24h | 7d) + explicit From/To datetime inputs.
Metric dropdown — every metric your sources have reported (CPU Usage %, Memory Usage, Container Restarts, etc.).
Interval picker (1m | 5m | 15m | 1h) — bucket size for aggregation. Smaller buckets = more detail but noisier; larger = smoother but coarser.
Label dropdowns — container_name, deployment, namespace, node, pod, workload_kind. Each defaults to All. Pick specific values to narrow the plot.
Query button — runs the plot.

After clicking Query, the chart renders below the controls.

Top 10 series by value are plotted. If your filter matches more, the page tells you ("Showing top 10 of 368 series"). Tighten filters to bring the count down.
The legend below shows each plotted series with its label values, color-coded.
The Y-axis is shared across series (single scale). The X-axis is your time range.

Top-right of the page:

Refresh — re-runs the query with current filters. The page does not auto-refresh.
Save as Alert — opens the New Alert flow pre-populated with your current metric + scope filters. The threshold is yours to set in the alert form. See Authoring rules.

Always set the time range before picking a metric — narrowing it first makes the metric dropdown lookup faster.
For high-cardinality metrics, the "top 10 of N" cap is generally fine; pick a namespace or pod filter if you need to see a specific series that's not in the top 10.
The Explorer view doesn't persist. To save a plot, switch to Dashboards and add a Time Series widget configured the same way.