Skip to content
Get packs

Guide · Agent skill

How to use the Cloud and Kubernetes Security Assessment agent skill

Updated · written with AI from the project's own documentation and checked against our catalogue data

The cloud-k8s skill gives an AI agent a structured checklist for authorized security testing of cloud accounts, containers and Kubernetes clusters. It is aimed at security engineers reviewing their own environment, penetration testers with written scope, and platform teams auditing RBAC and container privileges. After this guide you will know what the skill does, how to install it, how to ask for it in plain language, and what it needs before it runs.

Cloud and Kubernetes Security Assessmentzhaoxuya520/reverse-skill · MIT

Quick start

  1. 1Download and unzip. Download the cloud-k8s ZIP from this page and unzip it so the cloud-k8s folder contains SKILL.md.
  2. 2Place or upload the skill. Copy the folder to ~/.claude/skills/ or a project's .claude/skills/, or upload it under Settings → Capabilities → Skills in the Claude apps.
  3. 3Install the external tools. Install kubectl, your cloud CLI, Trivy, kube-bench, kubeaudit, pacu, ScoutSuite and nuclei, since most are manual installs.
  4. 4Configure authorized credentials. Set up cloud access keys or Kubernetes service account credentials for the account or cluster you are authorized to test.
  5. 5Write a scoped request. Ask for the task in plain language and name the single account, cluster or namespace in scope.
  6. 6Review findings. Check each finding for reproduction steps and impact, and confirm no destructive actions were taken.

What the cloud-k8s skill makes the agent do

An agent skill is a folder of instructions that an AI agent loads when your request matches the skill's purpose. The cloud-k8s skill is described as covering authorized cloud, container and Kubernetes security assessment, including metadata SSRF, IAM misconfiguration, container escape paths and cluster RBAC review.

The agent works through the checklist in a fixed order. It starts with identity and scope, then checks cloud configuration such as public storage, metadata service exposure and IAM permissions. Next it reviews container risks like privileged settings, host mounts and capabilities, and finally Kubernetes RBAC, secrets and network policy.

  • Identity and scope: which credentials you hold and which account, cluster or namespace is in bounds
  • Cloud configuration: public storage buckets, IMDSv1 versus IMDSv2, role permissions
  • Containers: privileged mode, hostPath and hostNetwork, Linux capabilities, writable host paths, image CVEs with Trivy
  • Kubernetes: service account tokens, cluster-admin bindings, exposed secrets, admission webhooks, network policies

Install the cloud-k8s skill

Download the ZIP for the cloud-k8s skill from this page and unzip it. For a personal setup, place the skill folder in ~/.claude/skills/. For a single project, place it in that project's .claude/skills/ folder instead.

If you use the Claude apps, you can instead upload the skill under Settings → Capabilities → Skills.

  • Personal: ~/.claude/skills/cloud-k8s
  • Project only: .claude/skills/cloud-k8s inside the project folder
  • Claude apps: Settings → Capabilities → Skills, then upload

Install the tools the skill expects

The skill relies on external command-line tools. Most must be installed manually, so check them before your first run. The skill also needs credentials for the account you are authorized to test.

Required tools named in the skill: kubectl, the AWS, gcloud or Azure CLI (configured with credentials for the authorized account), Trivy, kube-bench, kubeaudit, pacu, ScoutSuite and nuclei. The skill's tool table marks kubectl, kube-bench, kubeaudit, pacu and ScoutSuite as manual installs, and says Trivy and nuclei can be bootstrapped if available.

  • Cloud or Kubernetes credentials for the authorized account (access keys or service account credentials you supply)
  • kubectl and your cloud provider CLI, configured for that account
  • Trivy, kube-bench, kubeaudit, pacu, ScoutSuite and nuclei

Trigger the skill with natural requests

You do not need a special command. Describe the authorized task in plain language and name the scope. The skill's own first step is to confirm the target is cloud metadata, container, Kubernetes or IAM work, not a general web scan. Web application scanning is routed to a different toolset.

Include the boundary in your request, such as a single account, a single cluster or a single namespace. The skill asks the agent to start from identity and exposure and not to scan broadly by default.

  • "Using the cloud-k8s skill, review RBAC in namespace payments on our staging cluster. I have written authorization for this cluster."
  • "Check whether our AWS account's instance metadata is set to IMDSv2 and whether any public S3 buckets exist in account 1234 that I own."
  • "Run the container escape checklist against the api deployment in our own cluster and list privileged, hostPath and capability risks."

Example workflow on your own cluster

A typical review starts by confirming identity. The skill's checklist uses commands such as the ones below to show what the current credentials can do. Replace them with the equivalents for your provider and only run them inside the account you are authorized to test.

The checklist then asks for container settings to be checked, such as whether containers run as root or mount docker.sock, and for cluster bindings to be reviewed for over-broad cluster-admin grants.

kubectl auth can-i --list kubectl get clusterrolebindings

Safety, scope and network use

The skill is written to be used only against systems you are explicitly authorized to test. It asks the agent to avoid destructive actions and broad network scanning. Testing cloud accounts, clusters or tenants you do not own or lack written permission to assess is illegal, and the skill explicitly prohibits it.

The skill runs scripts and uses the network. Its reference checklist and SKILL.md are plain text, so read them before first use. Keep the agent's network access limited to the in-scope targets you name.

  • Confirm written authorization and the exact account, cluster or namespace before starting
  • Ask the agent to avoid destructive actions such as deleting resources or changing permissions
  • Do not let the agent scan other tenants or unrelated networks

Tips for better results

Give the agent the scope and credentials in the first message. Ask for findings to include reproduction steps and impact, since the skill's self-check expects both. Keep the checklist file open so you can compare the agent's output against the items it is meant to cover.

  • State the scope (single account, cluster or namespace) every time
  • Ask for reproduction steps and impact for each finding
  • Use the reference checklist to verify coverage of IMDS, Kubernetes high-risk items and container items

What people use it for

  • Reviewing cluster RBAC and flagging over-broad cluster-admin bindings in your own Kubernetes cluster
  • Checking whether your cloud instances enforce IMDSv2 and whether metadata is reachable from a web app's SSRF path
  • Auditing a container deployment for privileged mode, hostPath mounts, dangerous capabilities or a mounted docker.sock
  • Checking your own cloud account for public storage buckets and overly broad IAM permissions
  • Preparing a written cloud and container section for a penetration test report under a signed scope

Frequently asked questions

How do I install the cloud-k8s agent skill?

Download the ZIP from this page and unzip it into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder. In the Claude apps you can instead upload it under Settings → Capabilities → Skills.

What do I need before running cloud-k8s?

You need kubectl, your AWS, gcloud or Azure CLI configured with credentials for the authorized account, and the tools Trivy, kube-bench, kubeaudit, pacu, ScoutSuite and nuclei. Most of these must be installed manually. You also need cloud access keys or Kubernetes service account credentials from the user.

Is cloud-k8s free to use?

The source does not state pricing or licence terms beyond the repository licence, which is MIT. Check the repository's licence file for the exact terms before using it.

Can I use cloud-k8s to test a cloud account I do not own?

No. The skill is written for systems you are explicitly authorized to test, and it explicitly prohibits testing cloud accounts, clusters or tenants you do not own or have written permission to assess.

Does cloud-k8s scan web applications?

Not as its main purpose. The skill says general web application scanning belongs to a different tool set, so the agent is meant to route ordinary web scans elsewhere.

Is cloud-k8s suitable for beginners?

The skill is best suited to people with security background, because its output can describe exploitable paths that need expert judgment to handle safely.

New agent skills, repos and AI tools — every week

The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.

Our verdict

Use cloud-k8s if you are a security engineer, penetration tester with written scope, or platform team reviewing your own cloud, container and Kubernetes setup. It gives the agent a disciplined checklist that starts with scope and avoids destructive or broad scanning. If you lack security background or need general web application testing, choose a different tool.

Download the skill