Summary
Fission allowed low-privilege developers who could create KubernetesWatchTriggers (KWT) in their own namespace to observe events occurring in other namespaces.
Under misconfigured authorization boundaries, this cross-namespace watch capability can enable persistent surveillance of resources and activity outside the attacker's intended scope.
Affected context
The issue occurs in Fission deployments where KubernetesWatchTriggers are used to react to Kubernetes events.
It is relevant when developers are delegated limited rights to create KWT objects in specific namespaces, but the Fission control plane or its backing Kubernetes roles do not strictly enforce namespace scoping for the underlying watch operations.
In such cases, a trigger defined in one namespace can still be bound to events from other namespaces, creating an unintended cross-namespace observation channel.
Exploit preconditions
- Attack surface: Fission's KubernetesWatchTrigger (KWT) mechanism exposed via the Fission API and underlying Kubernetes resources.
- Required access: Ability for a low-privilege user to create or configure a KWT in at least one namespace they control.
- Trigger point: The moment the Fission system processes the KWT and establishes watch streams against the Kubernetes API.
- Why exploitation is possible: The authorization checks and/or RBAC bindings governing which namespaces can be watched are not properly restricted, allowing a trigger scoped to one namespace to subscribe to events in other namespaces.
Why this happens
The root cause is an authorization and access control design flaw where watch operations initiated by a KWT are not correctly constrained to the creator's namespace.
This maps to CWE-284: Improper Access Control and CWE-862: Missing Authorization, as the system fails to enforce per-namespace authorization for the target of the watch.
Conceptually, a higher-privilege service account or controller performs cross-namespace operations on behalf of untrusted users without re-checking whether the requested namespaces are within the caller's permitted scope.
Potential impact
- An attacker may passively monitor resource changes in other namespaces.
- An attacker could infer deployment activity, scale events, and operational patterns of other teams or tenants.
- An attacker can collect metadata about objects, labels, and configurations that may aid in further reconnaissance or lateral movement.
- Cross-tenant or cross-team information boundaries can be weakened in multi-tenant clusters.
General mitigation guidance
Mitigation generally focuses on tightening authorization boundaries and eliminating unintended privilege delegation.
Common defensive patterns include:
- Enforcing strict namespace scoping in all watch and list operations initiated by triggers.
- Ensuring that any controller or service account acting on user-defined triggers re-validates the requested namespace against the user's permissions.
- Designing trigger definitions so they cannot reference resources or namespaces outside the caller's scope by default.
- Applying least-privilege RBAC for Fission components, avoiding cluster-wide watch permissions where per-namespace rights are sufficient.
- Segmenting tenants and teams logically and, where applicable, physically to reduce cross-namespace visibility.
How teams detect issues like this
High-level detection approaches may include:
- Static review of Fission controller and trigger code paths to identify where namespace parameters are accepted and whether authorization is re-enforced at that boundary.
- RBAC policy review to ensure that service accounts backing KWTs are limited to the minimal namespaces and resource types required.
- Audit log analysis of Kubernetes API server events to detect unexpected watch or list operations targeting namespaces not associated with the requesting user or team.
- Use of policy-as-code tools (such as admission controllers or OPA-based policies) to prevent creation of KWTs that specify disallowed namespaces or overly broad selectors.

