菜单
Use when the user wants to uninstall, remove, or clean up the quanton-operator Helm chart and its leftover secrets from a Kubernetes cluster
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
| name | clean-uninstall |
| description | Use when the user wants to uninstall, remove, or clean up the quanton-operator Helm chart and its leftover secrets from a Kubernetes cluster |
| allowed-tools | Bash, Read, Glob, Grep, AskUserQuestion |
Fully removes the Quanton Operator Helm release, cleans up secrets it created across namespaces, and removes the CRD.
helm uninstall removes the chart resources but leaves behind:
quanton-operator-certquanton-operator-docker-secretquanton-operator-mtls-secretquantonsparkapplications.quantonsparkoperator.onehouse.ai) — Helm never deletes CRDs on uninstall by design.The cleanup script at scripts/cleanup-secrets.sh handles finding and deleting secrets and the CRD.
Check that kubectl and helm are available. Stop if either is missing.
Run these and show the user what's installed:
helm list -A | grep quanton-operator
kubectl get secrets -A | grep quanton-operator
kubectl get crd quantonsparkapplications.quantonsparkoperator.onehouse.ai 2>/dev/null
If neither the Helm release, secrets, nor CRD are found, tell the user there's nothing to clean up and stop.
Show what will be removed (Helm release and/or secrets) and ask the user to confirm before proceeding.
If the Helm release exists, uninstall it:
helm uninstall quanton-operator -n quanton-operator
Wait for completion. If the namespace quanton-operator is now empty, ask the user if they'd like to delete it too.
Run the cleanup script in dry-run mode first to show what will be deleted:
./scripts/cleanup-secrets.sh
Then run with --confirm:
./scripts/cleanup-secrets.sh --confirm
Run these to confirm everything is gone:
kubectl get secrets -A | grep quanton-operator
kubectl get crd quantonsparkapplications.quantonsparkoperator.onehouse.ai 2>/dev/null
Report the result to the user.
helm uninstall fails, show the error and check if the release even exists (helm list -A).chmod +x scripts/cleanup-secrets.sh first.Run the TPC-DS benchmark comparing OSS Apache Spark vs Quanton on Kubernetes, with interactive configuration and live progress updates. Optionally enables the in-driver Spark Agent (sidebar with Chat, Recommendations, Diagnostics, Monitor, Cost, SHS cohort) on the Quanton run so the user can interact with it live while the benchmark executes.
Run the Hudi or Iceberg MERGE INTO demo on minikube — creates a customers table, inserts 10 rows, runs MERGE INTO with 3 updates + 3 inserts, and verifies the final state (10 -> 13 rows, 3 'vip')
Run the Hudi or Iceberg clustering demo on minikube — writes a 100-row table with a complex (Struct/Array/Map/nested) schema, forces a many-tiny-files layout, then triggers the format's native clustering procedure with spark.quanton.clustering.accelerate=true and verifies it succeeded
Set up minikube, install Spark Operator and Quanton Operator, and run an example Quanton Spark job to demo that everything works