Features
Cluster insight
The questions that are not about a single object.
What is broken right now
A ranked queue of live failures, rather than a list you read looking for red.
Cluster overview
Version, node readiness, allocatable against usage, pods by phase, recent warnings.
Live usage in the tables
CPU and memory from metrics-server, in the list itself.
Usage history
CPU and memory over time, not only the current sample.
Best-practice checks
A scan for the usual security, reliability and waste problems, leading with what it found. Findings from another scanner can be imported alongside.
128 checks, with baselines, mutes, custom CEL rules and imported scanner findings
Posture by framework control
CIS Kubernetes Benchmark, PSS and NSA/CISA controls, each with what fails it, or why nothing reading a live cluster can answer it.
Findings as CSV, JSON or SARIF
SARIF that existing tooling reads unchanged, with a fingerprint per finding so a later run deduplicates against an earlier one.
Checks on a timer
Re-run on an interval, keep what each run found, and post a webhook when posture moves.
Reserved against used
Requests per namespace and workload against measured usage, ranked by what could be given back.
RBAC visibility
Who can do what across the cluster, resolved from roles and bindings.
Admin-only in cluster mode
A cluster that stops answering
Named in a banner with how long it has been quiet and why, rather than reading as healthy.
Topology graph
Resource relationships drawn as a graph.
Event timeline
Events and resource changes on one time axis.
Resource changes are recorded; Kubernetes events stay separate
Service traffic map
Live traffic between services, read from a mesh or an eBPF source.
Cilium Hubble flow metrics through Prometheus
Upgrade impact
What a control-plane minor bump would break.
MCP server for AI agents
An agent queries the cluster through the tool.
A separate binary with 21 tools and 2 resources
Being built
- Reachability diagnosis
- Cost
- Image filesystem browser