Gateway-Fabric - CraftedSignal Threat Feedhttps://feed.craftedsignal.io/tags/gateway-fabric/Trending threats, MITRE ATT&CK coverage, and detection metadata. Fed continuously.Hugoenhello@craftedsignal.iohello@craftedsignal.ioWed, 14 Oct 2026 16:38:30 +0000https://feed.craftedsignal.io/favicon-32x32.pngCraftedSignal Threat Feedhttps://feed.craftedsignal.io/3232https://feed.craftedsignal.io/favicon.svgConfiguration Injection in NGINX Gateway Fabric via AuthenticationFilter CRDhttps://feed.craftedsignal.io/briefs/2026-10-nginx-gateway-fabric-cve-2026-91184/Wed, 14 Oct 2026 16:38:30 +0000hello@craftedsignal.iohttps://feed.craftedsignal.io/briefs/2026-10-nginx-gateway-fabric-cve-2026-91184/NGINX Gateway Fabric contains a configuration injection vulnerability in the AuthenticationFilter Custom Resource Definition that allows authenticated attackers to escape directive contexts and manipulate the shared NGINX data-plane configuration.NGINX Gateway Fabric is affected by a configuration injection vulnerability, tracked as CVE-2026-91184, involving the improper handling of the spec.basic.realm and spec.jwt.realm fields within the AuthenticationFilter Custom Resource Definition (CRD). When these fields are processed, the Gateway Fabric renders the input verbatim and without validation into the shared nginx.conf file used by the data plane. An attacker with the ability to define or modify an AuthenticationFilter CRD associated with an HTTPRoute or GRPCRoute on a shared Gateway can supply a malicious string. This string can close the existing auth_basic directive and the enclosing location block, enabling the injection of arbitrary NGINX configuration directives. This represents a significant control-plane security risk in multi-tenant Kubernetes environments.

Impact

An authenticated attacker capable of modifying AuthenticationFilter resources can compromise the shared gateway's data-plane configuration. This success could result in severe cross-tenant impact, including the unauthorized reading of co-located tenant secrets, Server-Side Request Forgery (SSRF) via the proxy_pass directive, theft of pod ServiceAccount tokens, and the hijacking of traffic destined for other tenants. The vulnerability is restricted to the control plane and does not involve direct data-plane exposure, but the configuration injection grants the attacker influence over the entire gateway's routing and security posture.

Recommendation

  • Monitor Kubernetes API audit logs for suspicious modifications to AuthenticationFilter custom resources, specifically looking for unusual characters or NGINX directive keywords within the realm field.
  • Implement and enforce Kubernetes Admission Controllers (e.g., OPA Gatekeeper or Kyverno) to validate AuthenticationFilter specs and reject input containing characters such as double quotes, curly braces, or semicolon characters in the realm fields.
  • Patch NGINX Gateway Fabric to the version that implements strict input validation and escaping for CRD fields.
]]>highadvisoryconfiguration-injectionkubernetesgateway-fabriccve