Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when working with WizTelemetry Auditing extension for KubeSphere, including installation, configuration, and audit query API
.claude/skills/kubesphere-whizard-auditing/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-17 | ✗→✓ | ▲ Improved | 91% | 0% |
| case-01 | ✗→✓ | ▲ Improved | 76% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 136% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 167% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 36% | 0% |
WizTelemetry Auditing is an extension component in the KubeSphere Observability Platform for Kubernetes and KubeSphere audit event collection, processing, and storage.
| Component | Description | Default Enabled | |-----------|-------------|-----------------| | kube-auditing | Kubernetes audit event collection and export | true |
REQUIRED: Complete all steps in order before generating InstallPlan.
⚠️ CRITICAL: DO NOT proceed until target clusters are determined.
Step 1.1: Get available clusters
bashkubectl get clusters -o jsonpath='{.items[*].metadata.name}'
Step 1.2: Determine target clusters
Ask user (if not specified):
Available clusters: host, dev
Which clusters do you want to deploy WizTelemetry Auditing to?MUST do this to get the latest version:
bashkubectl get extensionversions -l kubesphere.io/extension-ref=whizard-auditing -o jsonpath='{range .items[*]}{.spec.version}{"\n"}{end}' | sort -V | tail -1
This outputs the latest version (e.g., 1.4.0). Note this down - you'll use it in the InstallPlan.
⚠️ IMPORTANT: Complete prerequisite steps BEFORE this step.
Based on your selections:
⚠️ CRITICAL: InstallPlan metadata.name MUST be whizard-auditing. DO NOT use any other name.
⚠️ CRITICAL: config field is YAML format. You MUST:
⚠️ CRITICAL: All placeholders MUST be replaced with actual values. DO NOT leave them as placeholders.
yamlapiVersion: kubesphere.io/v1alpha1 kind: InstallPlan metadata: name: whizard-auditing spec: extension: name: whizard-auditing version: <VERSION> # From Step 2 enabled: true upgradeStrategy: Manual clusterScheduling: placement: clusters: - <TARGET_CLUSTERS>
Replace placeholders:
<VERSION>: From Step 2 (e.g., 1.4.0)<TARGET_CLUSTERS>: User-confirmed cluster namesNote: OpenSearch sink configuration (endpoints, auth) is provided by the vector extension. Make sure vector is installed and configured with OpenSearch before installing auditing.
To enable Doris sink for audit storage:
yamlapiVersion: kubesphere.io/v1alpha1 kind: InstallPlan metadata: name: whizard-auditing spec: extension: name: whizard-auditing version: <VERSION> # From Step 2 enabled: true upgradeStrategy: Manual config: | kube-auditing: sinks: opensearch: enabled: true index: prefix: "{{ .cluster }}-auditing" timestring: "%Y.%m.%d" doris: enabled: true fe: <DORIS_FE> be: <DORIS_BE> table: partitionUnit: DAY retentionPartition: 7 replicationNum: 2 clusterScheduling: placement: clusters: - <TARGET_CLUSTERS>
yamlapiVersion: kubesphere.io/v1alpha1 kind: InstallPlan metadata: name: whizard-auditing spec: extension: name: whizard-auditing version: <VERSION> # From Step 2 enabled: true upgradeStrategy: Manual config: | kube-auditing: sinks: opensearch: enabled: true index: prefix: "{{ .cluster }}-auditing" timestring: "%Y.%m.%d" ism_policy: enable: true min_index_age: "7d" clusterScheduling: placement: clusters: - <TARGET_CLUSTERS>
| Parameter | Type | Default | Description | |-----------|------|---------|-------------| | kube-auditing.sinks.opensearch.enabled | bool | true | Enable OpenSearch sink | | kube-auditing.sinks.opensearch.index.prefix | string | "{{ .cluster }}-auditing" | Index prefix | | kube-auditing.sinks.opensearch.index.timestring | string | "%Y.%m.%d" | Index time format |
| Parameter | Type | Default | Description | |-----------|------|---------|-------------| | kube-auditing.sinks.doris.enabled | bool | false | Enable Doris sink | | kube-auditing.sinks.doris.fe | string | "" | Doris Frontend address | | kube-auditing.sinks.doris.be | string | "" | Doris Backend address | | kube-auditing.sinks.doris.table.partitionUnit | string | DAY | Partition unit | | kube-auditing.sinks.doris.table.retentionPartition | int | 7 | Retention partition | | kube-auditing.sinks.doris.table.replicationNum | int | 2 | Replication number |
| Parameter | Type | Default | Description | |-----------|------|---------|-------------| | kube-auditing.ism_policy.enable | bool | false | Enable Index State Management policy | | kube-auditing.ism_policy.min_index_age | string | "7d" | Minimum index retention period |
bashcurl -X GET "http://whizard-telemetry-apiserver.extension-whizard-telemetry.svc:80/kapis/logging.kubesphere.io/v1alpha2/auditing?operation=query&sort=desc&size=10&cluster=host" \ -H "X-Remote-User: admin"
Query Parameters:
| Parameter | Type | Default | Description | |-----------|------|---------|-------------| | operation | string | query | Operation type: query/statistics/histogram/export | | workspace_filter | string | | Comma-separated list of workspaces | | workspace_search | string | | Fuzzy match workspace names | | objectref_namespace_filter | string | | Comma-separated list of namespaces (ObjectRef.Namespace) | | objectref_namespace_search | string | | Fuzzy match namespace names | | objectref_name_filter | string | | Comma-separated list of object names | | objectref_name_search | string | | Fuzzy match object names | | level_filter | string | | Audit level: Metadata/Request/RequestResponse | | verb_filter | string | | Comma-separated list of verbs (create, update, delete, etc.) | | user_filter | string | | Comma-separated list of users | | user_search | string | | Fuzzy match username | | group_search | string | | Fuzzy match user groups | | source_ip_search | string | | Fuzzy match source IPs | | objectref_resource_filter | string | | Comma-separated list of resources | | objectref_subresource_filter | string | | Comma-separated list of subresources | | response_code_filter | string | | Comma-separated list of response codes | | response_status_filter | string | | Comma-separated list of response statuses | | start_time | string | | Start time (seconds since epoch) | | end_time | string | | End time (seconds since epoch) | | interval | string | 15m | Time interval for histogram | | sort | string | desc | Sort order: asc/desc | | from | int | 0 | Offset | | size | int | 10 | Number of results | | cluster | string | host | Cluster name |
bashkubectl get installplan whizard-auditing kubectl get extensionversions -l kubesphere.io/extension-ref=whizard-auditing
Uninstall from all clusters:
bashkubectl delete installplan whizard-auditing
Uninstall from specific cluster:
To remove WizTelemetry Auditing from a specific cluster, update the InstallPlan by removing that cluster from clusterScheduling.placement.clusters:
yamlapiVersion: kubesphere.io/v1alpha1 kind: InstallPlan metadata: name: whizard-auditing spec: extension: name: whizard-auditing version: <VERSION> enabled: true upgradeStrategy: Manual clusterScheduling: placement: clusters: - <REMAINING_CLUSTERS> # Remove the cluster you want to uninstall from
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-17 | fail→pass | 12,245 | 6,087 | -50% | 1 | 1 | 0% | 1,808 | 3,452 | +91% | 0 | 0 | — |
case-01 | fail→pass | 13,562 | 10,691 | -21% | 1 | 1 | 0% | 2,141 | 3,771 | +76% | 0 | 0 | — |
case-02 | fail→pass | 23,677 | 12,898 | -46% | 1 | 1 | 0% | 1,681 | 3,961 | +136% | 0 | 0 | — |
case-03 | fail→pass | 6,690 | 3,699 | -45% | 1 | 1 | 0% | 1,164 | 3,110 | +167% | 0 | 0 | — |
case-04 | fail→fail | 11,591 | 4,496 | -61% | 1 | 1 | 0% | 1,631 | 3,222 | +98% | 0 | 0 | — |
case-05 | fail→pass | 22,237 | 3,599 | -84% | 1 | 1 | 0% | 2,179 | 2,963 | +36% | 0 | 0 | — |
case-06 | fail→pass | 13,975 | 5,067 | -64% | 1 | 1 | 0% | 2,264 | 3,378 | +49% | 0 | 0 | — |
case-07 | fail→pass | 7,487 | 6,201 | -17% | 1 | 1 | 0% | 1,451 | 3,608 | +149% | 0 | 0 | — |
case-08 | fail→pass | 13,743 | 3,671 | -73% | 1 | 1 | 0% | 2,093 | 3,017 | +44% | 0 | 0 | — |
case-09 | fail→pass | 9,323 | 1,930 | -79% | 1 | 1 | 0% | 1,420 | 2,607 | +84% | 0 | 0 | — |
case-10 | fail→pass | 11,602 | 5,928 | -49% | 1 | 1 | 0% | 1,986 | 3,689 | +86% | 0 | 0 | — |
case-11 | fail→pass | 9,614 | 2,712 | -72% | 1 | 1 | 0% | 1,878 | 2,845 | +51% | 0 | 0 | — |
case-12 | pass→pass | 7,786 | 2,057 | -74% | 1 | 1 | 0% | 1,100 | 2,682 | +144% | 0 | 0 | — |
case-13 | fail→pass | 6,497 | 3,356 | -48% | 1 | 1 | 0% | 1,165 | 2,935 | +152% | 0 | 0 | — |
case-14 | fail→pass | 7,681 | 2,626 | -66% | 1 | 1 | 0% | 1,151 | 2,767 | +140% | 0 | 0 | — |
case-15 | fail→pass | 6,404 | 3,344 | -48% | 1 | 1 | 0% | 1,241 | 2,984 | +140% | 0 | 0 | — |
case-16 | fail→pass | 9,015 | 2,854 | -68% | 1 | 1 | 0% | 1,356 | 2,764 | +104% | 0 | 0 | — |
case-18 | fail→pass | 9,775 | 2,402 | -75% | 1 | 1 | 0% | 1,544 | 2,662 | +72% | 0 | 0 | — |
case-19 | fail→pass | 14,887 | 2,716 | -82% | 1 | 1 | 0% | 1,583 | 2,683 | +69% | 0 | 0 | — |
case-20 | pass→pass | 5,045 | 5,538 | +10% | 1 | 1 | 0% | 892 | 3,250 | +264% | 0 | 0 | — |
case-21 | pass→pass | 5,058 | 5,154 | +2% | 1 | 1 | 0% | 895 | 3,284 | +267% | 0 | 0 | — |
case-22 | pass→pass | 5,070 | 5,919 | +17% | 1 | 1 | 0% | 757 | 3,210 | +324% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +77 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.