Google OSV-SCALIBR EUVD-2026-89370

Exploitation requires that OSV-SCALIBR (version 0.3.6 through 0.5.0) be invoked to scan an attacker-controlled VMDK disk image, and that the scan reach the embedded VMDK filesystem extractor for FAT32, ext4, or NTFS filesystems. The attacker must control the scan target content (the crafted image with directory entries whose names contain path-separator/traversal sequences); they do not need credentials on the host. Limiting factors: the attacker cannot trigger this remotely against the scanner service directly - a user or automation must actually run a scan against the malicious image (UI:P / AV:L), and impact severity scales with the scanner's process privileges (writing outside the destination directory is most dangerous when the scanner runs as root in CI/container pipelines). Versions before 0.3.6 or at/after the PR #2030 fix are not affected.

The CVSS 4.0 base score of 6.9 (medium) reflects a genuine but constrained risk profile. The vector CVSS:4.0/AV:L/AC:L/AT:N/PR:N/UI:P/VC:N/VI:H/VA:H is internally consistent with the description: the attack vector is Local (AV:L) because exploitation requires the target to actually scan an attacker-supplied VMDK image rather than reaching the scanner over a network, and UI:P captures the passive user/automation action of initiating a scan. Impact is integrity- and availability-focused (VI:H/VA:H, VC:N) since arbitrary file write outside the destination directory can overwrite host files or plant executables but does not directly disclose data. PR:N is defensible because no authentication to the scanner is required - the trust boundary is the untrusted scan target itself. No EPSS score, no CISA KEV listing, and no public exploit code are present in the input, so this is 'no public exploit identified at time of analysis.' The realistic elevation comes from deployment context: OSV-SCALIBR is a supply-chain/SBOM scanner often run in CI/CD and container-analysis pipelines, sometimes as root, where an attacker-controlled artifact (e.g., a malicious disk image submitted for scanning) is exactly the input the tool ingests. This is a real priority for organizations running the affected 0.3.6-0.5.0 versions in automated pipelines, but not a mass-exploitable remote flaw. Key limiting signal: the local vector and required scanning action mean exploitation is opportunistic against automated scanners, not broadly wormable.

An attacker crafts a malicious VMDK image whose embedded FAT32/ext4/NTFS filesystem contains directory entries with path-traversal sequences (names containing '/'), then gets it into a scanning pipeline - for example by submitting it to an automated malware/SBOM analysis service or committing it to a repository scanned in CI. When OSV-SCALIBR's embedded VMDK filesystem extractor unpacks the image, insufficient path validation lets extracted files escape the destination directory and write to arbitrary host locations. No public exploit code is identified at time of analysis, but the fix diff clearly reveals the missing '/'-containing-name filter, making the flaw trivially reproducible.