OpenShift oc CVE-2026-115424

Exploitation requires a victim to run the 'oc image extract' command against a container image the attacker controls; the malicious image must embed symbolic or hard links whose targets are absolute paths or point to sibling directories of the extraction destination, combined with a chained file entry that writes through those links. Limiting factors: it is not remotely triggerable (AV:L) and depends entirely on user interaction (UI:R) - the target must choose to extract the attacker's image - and the blast radius is bounded by the privileges of the user invoking oc, so a non-root operator cannot modify root-owned files. There is no indication a special non-default configuration is needed; the flaw is in oc's default extraction link-handling.

This is a medium-severity issue (CVSS 5.5) and the signals consistently point to a real-but-non-urgent risk rather than an emergency. The CVSS vector CVSS:3.1/AV:L/AC:L/PR:N/UI:R/S:U/C:N/I:H/A:N shows a local attack vector with required user interaction: the attacker cannot reach a victim over the network and must instead get a target to voluntarily run 'oc image extract' against an attacker-supplied image, which sharply limits mass exploitation. Impact is integrity-only (I:H, with C:N and A:N), consistent with arbitrary file write/modification rather than data theft or service disruption. PR:N indicates no attacker authentication is required to the target system - the attacker's leverage is supplying a malicious image, not holding an account - so per the vector this is best described as unauthenticated-but-interaction-dependent. The vuln.today Decision label is 'this_month' (not act_now or emergency), and the issue is not listed in CISA KEV. EPSS is NOT provided in the input, so exploitation probability cannot be quantified here; no POC or exploit reference is present either. Net: a legitimate patch-this-month item for OpenShift users who run image extraction, not a drop-everything network-facing crisis, with the caveat that EPSS/KEV/POC data is simply absent rather than confirmed-benign.

An attacker publishes a malicious container image to a registry and induces a developer or operator to run 'oc image extract' against it. Because oc does not sanitize symbolic/hard link targets, links (combined with a chained file entry) that point to absolute paths or sibling directories cause the extraction to write attacker-controlled content outside the intended output directory, modifying files on the host with the privileges of the user running oc. No public exploit identified at time of analysis.