# Kubernetes events, a Coding Agents refresh, and richer dashboards

> Kubernetes events land inside Discover, Coding Agents monitoring gets cost-per-line and week-over-week views, plus unlisted dashboards and log variables.

Source: https://last9.io/changelog/kubernetes-events-a-coding-agents-refresh-and-richer-dashboards/

### See Kubernetes events without leaving Last9

The Kubernetes events your cluster emits now have a home in the product. Discover → Kubernetes gets an Events tab on cluster, node, pod, deployment, and other entity pages, plus a cluster-wide event feed. No more needing to know that events are stored as logs, or hunting for the right attribute names to find them.

### A bigger upgrade to Coding Agents monitoring

- Cost-per-line column on the Developers tab, with outlier markers, sorted by cost-per-line by default
- A "Top 3 outliers" widget on the Overview tab showing the most and least efficient developers
- Week-over-week comparison toggle, with delta percentages on Overview cards and a delta column on Developers
- Team filter chip so multi-team orgs can slice every panel by team
- A developer detail page for drilling into a single developer's cost, lines, and sessions
- A "Compare to..." button on each developer row, plus a side-by-side comparison view
- The cost-over-time chart now splits subagent activity out from the main thread instead of merging it into a single bar

### Improvements

- **Dashboards**
  - On-demand dashboards for diagnosis: when you investigate with the AI Assistant or MCP, it can build a dashboard on the fly and hand you a link. These are throwaway by design — accessed by URL, auto-expiring, and kept out of your saved dashboards index
  - Variables in log-based dashboards: define a variable and reference it as `$name` inside log panel queries
  - Stat (single-value) panels can fill the whole tile with a threshold color, not just the value text or a thin left border
  - Click on table panel rows now to drill into the operation and its traces
  - Multi-series line chart tooltips now show every series' value for a timestamp, so you can compare data points side by side instead of reading them one line at a time
- **Discover Services**: Upstream and downstream service maps now work for Rails services in production
- **Alerting**: The Alert Rule editor preview has a configurable time window (presets stay now-relative and re-anchor on refresh; absolute ranges pin exact bounds) and a clearer no-data state

### Fixes

- **Discover Services**: The shared crosshair across Overview charts disappeared when hovering over the right half of the Apdex Score and Response Time panels in some situations
- **Logs**
  - The range logs chart crashed with a render error when changing the time range on editor-mode aggregate queries
  - Log preview was broken for scheduled search
- **Synthetic Monitoring**
  - Comma-separated tags couldn't be added when creating a check
  - Custom HTTP headers were stored as `[object Object]`, and sensitive header values such as Authorization were returned in plaintext instead of being masked