CVE-2026-50564: Fission Environment CRD PodSpec Passthrough Enables Node Escape (CVSS 9.9)

What the Vulnerability Is

Fission's Environment Custom Resource exposes a podspec passthrough mechanism (poolspec.container / poolspec.podspec) that allows arbitrary Kubernetes PodSpec fields to be merged into the builder and executor pods created by the fission-builder and fission-router controllers. Because this passthrough is not restricted to a safe subset of fields, a principal can inject sensitive PodSpec keys such as hostPID, hostNetwork, hostIPC, and privileged securityContext settings. The root cause is a missing authorization/field-allowlist boundary (CWE-269, CWE-284) combined with an unsafe default trust assumption (CWE-693): the controllers trust that anyone able to write an Environment CRD is equally trusted to define raw pod security semantics. In multi-tenant Fission deployments, low-privileged function developers frequently have Environment CRD create/update rights, turning what should be a scoped function-runtime definition into a direct path to a privileged, host-namespace-sharing pod — and from there, to the underlying kubelet, container runtime socket, or co-located pods on the node.

Affected Software

• Vendor: go (github.com/fission/fission)
• Product: Fission serverless framework
• Affected versions: ≤ 1.23.0
• Patched version: not yet published
• CVSS: 9.9
• CWE: CWE-269 (Improper Privilege Management), CWE-284 (Improper Access Control), CWE-693 (Protection Mechanism Failure)

Exploitation Status

A public proof-of-concept exists for this vulnerability (exploit status: poc-public). It is not currently listed as a known exploited vulnerability (KEV), and no evidence of in-the-wild weaponization is indicated in this record. That said, PoC availability against a CVSS 9.9 privilege-escalation-to-node-compromise flaw materially raises risk for any organization running multi-tenant Fission clusters where Environment CRD authoring is delegated to application teams. Defenders should treat this as an urgent priority for detection and mitigating controls (e.g., admission policies restricting PodSpec fields) until an upstream patch lands.

How Our Detection Catches It

We ship coverage for this technique across Microsoft Sentinel (KQL), Splunk (SPL), Elastic (EQL), IBM QRadar (AQL), Sumo Logic, Google Chronicle (YARA-L), and CrowdStrike (CQL). At a high level, the detection logic focuses on the intersection of Fission control-plane activity and anomalous pod security posture:

• Monitoring Environment CRD create/update events for podspec passthrough fields (poolspec.container, poolspec.podspec) that set host-namespace flags or privileged security contexts.
• Correlating Kubernetes API server audit logs for pods spawned by fission-builder / fission-router service accounts that request hostPID, hostNetwork, hostIPC, or privileged: true.
• Flagging identities with Environment CRD write access that lack corresponding cluster-admin or node-access entitlements, surfacing the privilege mismatch that enables escalation.
• Alerting on subsequent host-level indicators consistent with a pivot off a privileged builder pod, such as unexpected access to the container runtime socket or kubelet API from a Fission-managed workload.

This layered approach lets defenders catch the escalation at the point of CRD manipulation, at pod admission/creation, or during the post-escalation pivot — regardless of which SIEM stack is in use.

Full Detection Details

For the complete KQL, SPL, and other SIEM queries, MITRE ATT&CK mapping, and remediation guidance, see the full detection page for CVE-2026-50564.