Woodpecker: Privilege escalation via unrestricted serviceAccountName in the Kubernetes backend

Impact

A privilege escalation vulnerability affects Woodpecker instances using the Kubernetes backend.

The pipeline option backend_options.kubernetes.serviceAccountName was passed directly to the pod spec without any admin gating.

Who is impacted: any operator running the Kubernetes backend. Any user with Push permission on a connected repository can run pipeline pods under an arbitrary ServiceAccount in the pipeline namespace, gaining that account's RBAC permissions. If a privileged ServiceAccount is reachable in that namespace, this can lead to secret exfiltration (database credentials, API keys, TLS certs) and full cluster takeover.

Patches

https://github.com/woodpecker-ci/woodpecker/pull/6792

Workarounds

Operators who cannot upgrade immediately can mitigate by any of:

- Restrict Push access on repositories connected to the Kubernetes-backed instance to trusted users only. - Harden the pipeline namespace: ensure no privileged ServiceAccount exists or is bound in the namespace where pipeline pods run; keep the default ServiceAccount minimally privileged. - Disable ServiceAccount token automounting for ServiceAccounts that should not be used by pipelines. - Enforce an admission policy (e.g. OPA/Gatekeeper, Kyverno, or a ValidatingAdmissionPolicy) that rejects pipeline pods setting an unexpected serviceAccountName. - Use a dedicated, isolated namespace per org/instance with no sensitive RBAC bindings.

Resources

- Vulnerable option introduced in commit 609ba481b5e912f59aaae8ca7bc22b44523c5e37 - Affected versions: v1.0.0 through v3.15.0 - Source: pipeline/backend/kubernetes/backend_options.go (field ServiceAccountName), pipeline/backend/kubernetes/pod.go (assigned to pod spec with no gating)