v1.4.31

Multi-Cluster Kubernetes

• Connect multiple Kubernetes clusters without an Enterprise license.
• Scan fleet cards showing health, Kubernetes version, node and Pod readiness,
  warnings, issues, and snapshot status.
• Open a cluster's dashboard, switch between clusters, and share direct links
  to its views and resources.
• Keep other clusters usable when one is unreachable, with a separate status
  and Retry control for the affected cluster.

Cluster-Targeted Investigations

• Select clusters in web chat and the CLI, and compare environments using
  read-only Kubernetes tools.
• Attach resources from the selected cluster and follow citations back to the
  correct dashboard.
• Keep approval-gated actions bound to their target cluster and separate
  action credentials.

Live Pod Logs

• Keep log streams tied to the selected cluster, Pod, and container.
• Cancel the previous stream when switching clusters, closing the drawer, or
  leaving the page.
• Report unavailable Previous-container logs clearly and recover to current
  logs without losing the viewer's controls.
• Retain pause/resume, timestamps, filtering, and scrubbed copy/download output.

Kubernetes Break and Crash Evidence

• Explain configuration, startup, image, scheduling, eviction, init-container,
  Job, node-health and probe-warning failures beyond OOMKilled.
• Count distinct affected pods rather than containers, show lifetime restarts
  separately, and distinguish actual termination signals from exit-code hints.
• Avoid false breaks when a healthy workload has leftover terminal pods.
• Keep captured crash history after pod deletion and Versus restarts, with
  paged observations and details beyond compact summaries.
• Distinguish observation windows from actual termination finish times. History
  defaults to 24 hours by observation time, subject to capacity limits; expose
  gaps and incomplete evidence without claiming every event was captured.

Enterprise

• Limit member and team infrastructure access to selected clusters while
  retaining existing role and MFA requirements.
• Hide unauthorized clusters consistently across fleet cards, dashboards,
  logs, attachments, and investigation tools.
• Preserve cluster restrictions during role changes.
• Apply cluster access checks to action proposal visibility, approval, and
  rejection.
• Keep traffic data associated with its explicitly configured cluster.
• Record new timeline changes with the AI agent disabled and support
  Kubernetes resource attachments in chat.

Configuration

Existing single-cluster configuration and dashboard behavior remain unchanged.
Multi-cluster is opt-in: set connectors.kubernetes.multiple: true and place
each cluster's connection settings under connectors.kubernetes.clusters.
Each entry needs a unique cluster_id; an optional display_name labels its
fleet card. Cluster settings do not inherit top-level connection credentials.

Known Limitation

Preserving existing OSS Kubernetes timeline, approval, and chat history after
an Enterprise upgrade is planned but not included in this release.

What's Changed

• feat(kubernetes): add opt-in multi-cluster fleet and investigations by @hoalongnatsu in #383
• feat(agent): support history and update docs for Kubernetes by @hoalongnatsu in #384
• feat(kubernetes): add break timelines and continuous change capture by @hoalongnatsu in #385
• feat(kubernetes): expand break evidence and retain crash history by @hoalongnatsu in #386

Full Changelog: v1.4.30...v1.4.31