OpenShift source-to-image CVE-2026-107174

Exploitation requires that the attacker can introduce a malicious builder image into a source-to-image (S2I) build and that S2I unpacks that image's archive files, which do not sanitize symbolic links resolving to absolute paths outside the extraction directory (CWE-61). The limiting factors are significant: the attacker must have the privilege or access to supply or reference the builder image used by a build (consistent with CVSS PR:L), so this is not an unauthenticated drive-by. There is no required end-user interaction (UI:N), but exploitation is confined to environments that run S2I builds on untrusted or attacker-influenced builder images; deployments that only use trusted, vetted builder images are not realistically exposed.

This is a moderate-severity, conditionally-exploitable issue rather than an urgent priority. Red Hat's CVSS:3.1 vector (AV:N/AC:L/PR:L/UI:N/S:C/C:L/I:L/A:N, base 6.4) indicates a network-reachable flaw requiring low privileges, with a scope change (S:C) reflecting escape from the build sandbox to the host, but only low confidentiality and integrity impact and no availability impact. The 'network' attack vector is somewhat generous: the real precondition is that an attacker can supply a malicious builder image into an S2I build, which in practice means possessing at least developer-level build privileges or control over a referenced image registry - consistent with PR:L. No EPSS score, CISA KEV listing, or public POC is present in the provided data, so exploitation signals are entirely absent beyond the vendor-assigned CVSS. There is no conflict among the signals; rather, there is a scarcity of them. Net assessment: a legitimate sandbox-escape/path-traversal weakness worth patching in CI/CD build infrastructure, but not a drop-everything emergency given the privilege requirement, the low technical impact, and no evidence of exploitation.

An attacker with the ability to trigger an OpenShift source-to-image build supplies a malicious builder image containing archive entries with symbolic links pointing to absolute paths outside the extraction directory. When S2I unpacks the archive, the unsanitized symlinks allow writing to or reading from arbitrary host locations, escaping the build sandbox and enabling limited information disclosure or file modification on the host. No public exploit identified at time of analysis, so this reflects the described mechanism rather than a weaponized PoC.