Canonical LXD EUVD-2026-57540

Exploitation requires the attacker to be able to import or unpack an image archive into LXD (an image-management privilege, consistent with PR:L), and to supply a crafted archive whose metadata.yaml entry is a symbolic link pointing to a chosen host path. The exploit is gated by LXD processing that attacker-controlled archive during import/unpack; it does not require user interaction (UI:N) and is reachable over the LXD API network vector (AV:N). Limiting factors: the attacker must already possess image-import rights rather than being fully anonymous, so environments that forbid untrusted users from importing images are not exposed, and successful host impact assumes the LXD daemon runs with sufficient (typically root) privilege to reach the targeted paths.

The provided CVSS 3.1 vector (AV:N/AC:L/PR:L/UI:N/S:C/C:H/I:H/A:H, 9.9) reflects a genuinely severe primitive - network-reachable via the LXD API, low complexity, scope-changing, and full CIA impact on the host - but PR:L confirms the attacker must already hold some authenticated privilege to import images, so this is not an anonymous internet-wide bug. No EPSS score, KEV entry, POC, or SSVC decision data was supplied, so real-world exploitation likelihood cannot be quantified from the input; the absence of KEV/POC signals means priority rests on the intrinsic severity and the value of the target rather than observed mass exploitation. One consistency note: the reference tag classifies this only as 'Information Disclosure,' which understates the vulnerability - the description and the I:H/A:H metrics indicate arbitrary file WRITE, not just read, so defenders should treat it as an integrity/host-compromise issue, not merely disclosure. Given LXD daemons typically run as root, the write primitive plausibly escalates to full host takeover, making this a real priority wherever untrusted or semi-trusted users can push images.

A semi-trusted tenant or CI account that holds LXD image-import privileges crafts an image archive in which metadata.yaml is a symlink pointing to a sensitive host file (for example a systemd unit, an authorized_keys file, or /etc configuration). When LXD imports and unpacks the archive, it follows the symlink and reads or overwrites the targeted host path as root, enabling credential theft or code execution on the host. No public exploit code was identified at time of analysis, but the low attack complexity makes weaponization straightforward once the archive format is understood.