docs/getting-started.md
Kubescape can run as a command line tool on a client, as an operator inside a cluster, as part of your CI/CD process, or more.
The best way to get started with Kubescape is to download it to the machine you use to manage your Kubernetes cluster.
curl -s https://raw.githubusercontent.com/kubescape/kubescape/master/install.sh | /bin/bash
(We're a security product; please read the file before you run it!)
You can also check other installation methods
kubescape scan
You will see output like this:
Kubescape security posture overview for cluster: minikube
In this overview, Kubescape shows you a summary of your cluster security posture, including the number of users who can perform administrative actions. For each result greater than 0, you should evaluate its need, and then define an exception to allow it. This baseline can be used to detect drift in future.
Control plane
╭────┬─────────────────────────────────────┬──────────────────────────────────────────────╮
│ │ Control Name │ Docs │
├────┼─────────────────────────────────────┼──────────────────────────────────────────────┤
│ ✅ │ API server insecure port is enabled │ https://kubescape.io/docs/controls/c-0005/ │
│ ❌ │ Anonymous access enabled │ https://kubescape.io/docs/controls/c-0262/ │
│ ❌ │ Audit logs enabled │ https://kubescape.io/docs/controls/c-0067/ │
│ ✅ │ RBAC enabled │ https://kubescape.io/docs/controls/c-0088/ │
│ ❌ │ Secret/etcd encryption enabled │ https://kubescape.io/docs/controls/c-0066/ │
╰────┴─────────────────────────────────────┴──────────────────────────────────────────────╯
Access control
╭─────────────────────────────────────────────────┬───────────┬────────────────────────────────────╮
│ Control Name │ Resources │ View Details │
├─────────────────────────────────────────────────┼───────────┼────────────────────────────────────┤
│ Cluster-admin binding │ 1 │ $ kubescape scan control C-0035 -v │
│ Data Destruction │ 6 │ $ kubescape scan control C-0007 -v │
│ Exec into container │ 1 │ $ kubescape scan control C-0002 -v │
│ List Kubernetes secrets │ 6 │ $ kubescape scan control C-0015 -v │
│ Minimize access to create pods │ 2 │ $ kubescape scan control C-0188 -v │
│ Minimize wildcard use in Roles and ClusterRoles │ 1 │ $ kubescape scan control C-0187 -v │
│ Portforwarding privileges │ 1 │ $ kubescape scan control C-0063 -v │
│ Validate admission controller (mutating) │ 0 │ $ kubescape scan control C-0039 -v │
│ Validate admission controller (validating) │ 0 │ $ kubescape scan control C-0036 -v │
╰─────────────────────────────────────────────────┴───────────┴────────────────────────────────────╯
Secrets
╭─────────────────────────────────────────────────┬───────────┬────────────────────────────────────╮
│ Control Name │ Resources │ View Details │
├─────────────────────────────────────────────────┼───────────┼────────────────────────────────────┤
│ Applications credentials in configuration files │ 1 │ $ kubescape scan control C-0012 -v │
╰─────────────────────────────────────────────────┴───────────┴────────────────────────────────────╯
Network
╭────────────────────────┬───────────┬────────────────────────────────────╮
│ Control Name │ Resources │ View Details │
├────────────────────────┼───────────┼────────────────────────────────────┤
│ Missing network policy │ 13 │ $ kubescape scan control C-0260 -v │
╰────────────────────────┴───────────┴────────────────────────────────────╯
Workload
╭─────────────────────────┬───────────┬────────────────────────────────────╮
│ Control Name │ Resources │ View Details │
├─────────────────────────┼───────────┼────────────────────────────────────┤
│ Host PID/IPC privileges │ 2 │ $ kubescape scan control C-0038 -v │
│ HostNetwork access │ 1 │ $ kubescape scan control C-0041 -v │
│ HostPath mount │ 1 │ $ kubescape scan control C-0048 -v │
│ Non-root containers │ 6 │ $ kubescape scan control C-0013 -v │
│ Privileged container │ 1 │ $ kubescape scan control C-0057 -v │
╰─────────────────────────┴───────────┴────────────────────────────────────╯
Highest-stake workloads
────────────────────────
High-stakes workloads are defined as those which Kubescape estimates would have the highest impact if they were to be exploited.
1. namespace: gadget, name: gadget, kind: DaemonSet
'$ kubescape scan workload DaemonSet/gadget --namespace gadget'
2. namespace: kafka, name: my-cluster-kafka-0, kind: Pod
'$ kubescape scan workload Pod/my-cluster-kafka-0 --namespace kafka'
3. namespace: kafka, name: my-cluster-zookeeper-0, kind: Pod
'$ kubescape scan workload Pod/my-cluster-zookeeper-0 --namespace kafka'
Compliance Score
────────────────
The compliance score is calculated by multiplying control failures by the number of failures against supported compliance frameworks. Remediate controls, or configure your cluster baseline with exceptions, to improve this score.
* MITRE: 77.39%
* NSA: 69.97%
View a full compliance report by running '$ kubescape scan framework nsa' or '$ kubescape scan framework mitre'
What now?
─────────
* Run one of the suggested commands to learn more about a failed control failure
* Scan a workload with '$ kubescape scan workload' to see vulnerability information
* Install Kubescape in your cluster for continuous monitoring and a full vulnerability report: https://github.com/kubescape/helm-charts/tree/main/charts/kubescape-operator
Capabilities
Scan Kubernetes clusters, YAML files, Helm charts for misconfigurations. Kubescape will highlight the misconfigurations and provide remediation steps. The misconfigurations are based on multiple frameworks (including NSA-CISA, MITRE ATT&CK® and the CIS Benchmark).
kubescape scan
Scan a running Kubernetes cluster with the NSA framework:
kubescape scan framework nsa
Scan a running Kubernetes cluster with the MITRE ATT&CK® framework:
kubescape scan framework mitre
Scan for a specific control, using the control name or control ID. See the list of controls.
kubescape scan control c-0005 -v
kubescape scan --kubeconfig cluster.conf
kubescape scan --include-namespaces development,staging,production
kubescape scan --exclude-namespaces kube-system,kube-public
kubescape scan /path/to/directory-or-file
Take a look at the example.
Scan Kubernetes manifest files from a Git repository:
kubescape scan https://github.com/kubescape/kubescape
kubescape scan --exceptions examples/exceptions/exclude-kube-namespaces.json
Objects with exceptions will be presented as exclude and not fail.
See more examples about exceptions.
Kubescape automatically detects a Helm chart directory by the
presence of a Chart.yaml file.
kubescape scan /path/to/helm/chart/directory
Note
Kubescape will load the defaultvalues.yamlfile. To use a custom values file, use the--helm-valuesflag.
Kubescape automatically detects a Kustomize directory by the
presence of a kustomization.yaml, kustomization.yml, or kustomization file.
kubescape scan /path/to/kustomize/directory
Note
Kubescape will generate Kubernetes YAML objects using thekustomization.yamlfile and scan them for security. If a directory contains bothChart.yamlandkustomization.yaml, Kubescape will treat it as a Helm chart.
If the kubescape-operator is installed in your cluster, you can trigger scanning of the in cluster components from the kubescape CLI.
Trigger configuration scanning:
kubescape operator scan configurations
Trigger vulnerabilities scanning:
kubescape operator scan vulnerabilities
We offer two important metrics to assess compliance:
kubescape scan --compliance-threshold <SCORE_VALUE[float32]>
kubescape scan framework <FRAMEWORK_NAME> --compliance-threshold <SCORE_VALUE[float32]>
kubescape scan --format json --output results.json
kubescape scan --format junit --output results.xml
SARIF is a standard format for the output of static analysis tools. It is supported by many tools, including GitHub Code Scanning and Azure DevOps. Read more about SARIF.
kubescape scan --format sarif --output results.sarif
Note SARIF format is supported only when scanning local files or git repositories, but not when scanning a running cluster.
kubescape scan --format html --output results.html
It is possible to run Kubescape offline! Check out our video tutorial.
Download the controls and save them in the local directory. If no path is specified, they will be saved in ~/.kubescape.
kubescape download artifacts --output path/to/local/dir
Copy the downloaded artifacts to the offline system.
Scan using the downloaded artifacts:
kubescape scan --use-artifacts-from path/to/local/dir
You can also download a single artifact, and scan with the --use-from flag:
Download and save in a file. If no file name is specified, the artifact will be saved as ~/.kubescape/<framework name>.json.
kubescape download framework nsa --output /path/nsa.json
Copy the downloaded artifacts to the offline system.
Scan using the downloaded framework:
kubescape scan framework nsa --use-from /path/nsa.json
Kubescape can scan container images for vulnerabilities. It uses Grype to scan the images.
kubescape scan image nginx:1.19.6
kubescape scan image --username myuser --password mypassword myregistry/nginx:1.19.6
kubescape scan image nginx:1.19.6 -v
We publish a Helm chart for our in-cluster components. Please follow the instructions here
Scan your YAML files while writing them using our VS Code extension.
View Kubescape scan results directly in the Lens IDE using the Kubescape Lens extension.
Experiment with Kubescape in the Kubescape playground: this scenario will install a K3s cluster and Kubescape. You can start with any of the kubescape scan commands in the examples.
Kubescape can automatically fix misconfigurations found in your Kubernetes manifest files.
# First, scan and save results to JSON
kubescape scan /path/to/manifests --format json --output results.json
# Then apply fixes based on the scan results
kubescape fix results.json
| Flag | Description |
|---|---|
--dry-run | Preview changes without applying them |
--no-confirm | Apply fixes without confirmation prompts |
--skip-user-values | Skip changes that require user-defined values (default: true) |
# Preview fixes without applying
kubescape fix results.json --dry-run
# Apply fixes without prompts (useful for CI/CD)
kubescape fix results.json --no-confirm
Warning
The fix command modifies files in-place. Always review changes or use--dry-runfirst.
Kubescape can patch container images to fix OS-level vulnerabilities using Copacetic and BuildKit.
# Start buildkitd (if not already running)
sudo buildkitd &
# Patch an image
sudo kubescape patch --image docker.io/library/nginx:1.22
| Flag | Description | Default |
|---|---|---|
-i, --image | Image name to patch (required) | - |
-t, --tag | Tag for the patched image | <image>-patched |
-a, --addr | BuildKit daemon address | unix:///run/buildkit/buildkitd.sock |
--timeout | Patching timeout | 5m |
-u, --username | Registry username | - |
-p, --password | Registry password | - |
-v, --verbose | Show detailed output | false |
export BUILDKIT_VERSION=v0.11.4
export BUILDKIT_PORT=8888
# Start BuildKit in Docker
docker run --detach --rm --privileged \
-p 127.0.0.1:$BUILDKIT_PORT:$BUILDKIT_PORT/tcp \
--name buildkitd \
--entrypoint buildkitd \
"moby/buildkit:$BUILDKIT_VERSION" \
--addr tcp://0.0.0.0:$BUILDKIT_PORT
# Patch using TCP connection
kubescape patch -i nginx:1.22 -a tcp://0.0.0.0:$BUILDKIT_PORT
Note
Image patching can only fix OS-level vulnerabilities, not application-level ones.
For more details, see the Patch Command Documentation.
Kubescape can help manage Kubernetes Validating Admission Policies using CEL (Common Expression Language).
Install the Kubescape CEL admission policy library:
kubescape vap deploy-library | kubectl apply -f -
This deploys:
Bind policies to specific resources:
kubescape vap create-policy-binding \
--name my-policy-binding \
--policy c-0016 \
--namespace my-namespace | kubectl apply -f -
create-policy-binding| Flag | Description | Required |
|---|---|---|
-n, --name | Name of the policy binding | Yes |
-p, --policy | Policy/control to bind | Yes |
--namespace | Namespace selector (can be repeated) | No |
--label | Label selector in key=value format | No |
-a, --action | Action on failure: Deny, Audit, Warn | No (default: Deny) |
-r, --parameter-reference | Parameter reference object name | No |
# Create a policy that denies non-compliant resources in production
kubescape vap create-policy-binding \
--name deny-privileged-containers \
--policy c-0057 \
--namespace production \
--action Deny | kubectl apply -f -
Kubescape provides an MCP (Model Context Protocol) server for AI assistant integration, allowing natural language queries about your cluster's security posture.
kubescape mcpserver
The MCP server exposes these tools to AI assistants:
| Tool | Description |
|---|---|
list_vulnerability_manifests | Discover vulnerability scan results |
list_vulnerabilities_in_manifest | List CVEs in a specific manifest |
list_vulnerability_matches_for_cve | Get details for a specific CVE |
list_configuration_security_scan_manifests | List configuration scan results |
get_configuration_security_scan_manifest | Get configuration scan details |
Add to your Claude Desktop configuration:
{
"mcpServers": {
"kubescape": {
"command": "kubescape",
"args": ["mcpserver"]
}
}
}
For more details, see the MCP Server Documentation.
Manage Kubescape's cached configurations:
# View current configuration
kubescape config view
# Set account ID
kubescape config set accountID <your-account-id>
# Set cloud report URL
kubescape config set cloudReportURL <url>
# Delete cached configuration
kubescape config delete