/kube-hunter

Hunt for security weaknesses in Kubernetes clusters

Primary LanguagePythonApache License 2.0Apache-2.0

kube-hunter

GitHub Release Downloads Docker Pulls Build Status codecov Code style: black License Docker image

kube-hunter hunts for security weaknesses in Kubernetes clusters. The tool was developed to increase awareness and visibility for security issues in Kubernetes environments. You should NOT run kube-hunter on a Kubernetes cluster that you don't own!

Run kube-hunter: kube-hunter is available as a container (aquasec/kube-hunter), and we also offer a web site at kube-hunter.aquasec.com where you can register online to receive a token allowing you to see and share the results online. You can also run the Python code yourself as described below.

Explore vulnerabilities: The kube-hunter knowledge base includes articles about discoverable vulnerabilities and issues. When kube-hunter reports an issue, it will show its VID (Vulnerability ID) so you can look it up in the KB at https://aquasecurity.github.io/kube-hunter/

Contribute: We welcome contributions, especially new hunter modules that perform additional tests. If you would like to develop your modules please read Guidelines For Developing Your First kube-hunter Module.

kube-hunter demo video

Table of Contents

Hunting

Where should I run kube-hunter?

There are three different ways to run kube-hunter, each providing a different approach to detecting weaknesses in your cluster:

Run kube-hunter on any machine (including your laptop), select Remote scanning and give the IP address or domain name of your Kubernetes cluster. This will give you an attackers-eye-view of your Kubernetes setup.

You can run kube-hunter directly on a machine in the cluster, and select the option to probe all the local network interfaces.

You can also run kube-hunter in a pod within the cluster. This indicates how exposed your cluster would be if one of your application pods is compromised (through a software vulnerability, for example). (--pod flag)

Scanning options

First check for these pre-requisites.

By default, kube-hunter will open an interactive session, in which you will be able to select one of the following scan options. You can also specify the scan option manually from the command line. These are your options:

  1. Remote scanning

To specify remote machines for hunting, select option 1 or use the --remote option. Example: kube-hunter --remote some.node.com

  1. Interface scanning

To specify interface scanning, you can use the --interface option (this will scan all of the machine's network interfaces). Example: kube-hunter --interface

  1. Network scanning

To specify a specific CIDR to scan, use the --cidr option. Example: kube-hunter --cidr 192.168.0.0/24

  1. Kubernetes node auto-discovery

Set --k8s-auto-discover-nodes flag to query Kubernetes for all nodes in the cluster, and then attempt to scan them all. By default, it will use in-cluster config to connect to the Kubernetes API. If you'd like to use an explicit kubeconfig file, set --kubeconfig /location/of/kubeconfig/file.

Also note, that this is always done when using --pod mode.

Authentication

In order to mimic an attacker in it's early stages, kube-hunter requires no authentication for the hunt.

  • Impersonate - You can provide kube-hunter with a specific service account token to use when hunting by manually passing the JWT Bearer token of the service-account secret with the --service-account-token flag.

    Example:

    $ kube-hunter --active --service-account-token eyJhbGciOiJSUzI1Ni...
  • When runing with --pod flag, kube-hunter uses the service account token mounted inside the pod to authenticate to services it finds during the hunt.

    • if specified, --service-account-token flag takes priority when running as a pod

Active Hunting

Active hunting is an option in which kube-hunter will exploit vulnerabilities it finds, to explore for further vulnerabilities. The main difference between normal and active hunting is that a normal hunt will never change the state of the cluster, while active hunting can potentially do state-changing operations on the cluster, which could be harmful.

By default, kube-hunter does not do active hunting. To active hunt a cluster, use the --active flag. Example: kube-hunter --remote some.domain.com --active

List of tests

You can see the list of tests with the --list option: Example: kube-hunter --list

To see active hunting tests as well as passive: kube-hunter --list --active

Nodes Mapping

To see only a mapping of your nodes network, run with --mapping option. Example: kube-hunter --cidr 192.168.0.0/24 --mapping This will output all the Kubernetes nodes kube-hunter has found.

Output

To control logging, you can specify a log level, using the --log option. Example: kube-hunter --active --log WARNING Available log levels are:

  • DEBUG
  • INFO (default)
  • WARNING

Dispatching

By default, the report will be dispatched to stdout, but you can specify different methods by using the --dispatch option. Example: kube-hunter --report json --dispatch http Available dispatch methods are:

  • stdout (default)
  • http (to configure, set the following environment variables:)
    • KUBEHUNTER_HTTP_DISPATCH_URL (defaults to: https://localhost)
    • KUBEHUNTER_HTTP_DISPATCH_METHOD (defaults to: POST)

Advanced Usage

Azure Quick Scanning

When running as a Pod in an Azure or AWS environment, kube-hunter will fetch subnets from the Instance Metadata Service. Naturally this makes the discovery process take longer. To hardlimit subnet scanning to a /24 CIDR, use the --quick option.

Deployment

There are three methods for deploying kube-hunter:

On Machine

You can run kube-hunter directly on your machine.

Prerequisites

You will need the following installed:

  • python 3.x
  • pip
Install with pip

Install:

pip install kube-hunter

Run:

kube-hunter
Run from source

Clone the repository:

git clone https://github.com/aquasecurity/kube-hunter.git

Install module dependencies. (You may prefer to do this within a Virtual Environment)

cd ./kube-hunter
pip install -r requirements.txt

Run:

python3 kube_hunter

If you want to use pyinstaller/py2exe you need to first run the install_imports.py script.

Container

Aqua Security maintains a containerized version of kube-hunter at aquasec/kube-hunter. This container includes this source code, plus an additional (closed source) reporting plugin for uploading results into a report that can be viewed at kube-hunter.aquasec.com. Please note, that running the aquasec/kube-hunter container and uploading reports data are subject to additional terms and conditions.

The Dockerfile in this repository allows you to build a containerized version without the reporting plugin.

If you run kube-hunter container with the host network, it will be able to probe all the interfaces on the host:

docker run -it --rm --network host aquasec/kube-hunter

Note for Docker for Mac/Windows: Be aware that the "host" for Docker for Mac or Windows is the VM that Docker runs containers within. Therefore specifying --network host allows kube-hunter access to the network interfaces of that VM, rather than those of your machine. By default, kube-hunter runs in interactive mode. You can also specify the scanning option with the parameters described above e.g.

docker run --rm aquasec/kube-hunter --cidr 192.168.0.0/24

Pod

This option lets you discover what running a malicious container can do/discover on your cluster. This gives a perspective on what an attacker could do if they were able to compromise a pod, perhaps through a software vulnerability. This may reveal significantly more vulnerabilities.

The example job.yaml file defines a Job that will run kube-hunter in a pod, using default Kubernetes pod access settings. (You may wish to modify this definition, for example to run as a non-root user, or to run in a different namespace.)

  • Run the job with kubectl create -f ./job.yaml
  • Find the pod name with kubectl describe job kube-hunter
  • View the test results with kubectl logs <pod name>

Contribution

To read the contribution guidelines, Click here

License

This repository is available under the Apache License 2.0.