LibreDesk CVE-2026-111168

Exploitation requires (1) an authenticated LibreDesk agent account able to upload media via /api/v1/media, and (2) a crafted SVG whose Content-Type carries a trailing parameter (e.g. 'image/svg+xml; x') so the vulnerable code's exact string comparison against 'image/svg+xml' fails and the file is served inline instead of sandboxed; and (3) a victim agent or administrator who subsequently opens GET /uploads/{uuid} in their browser. Limiting factors that reduce real-world risk: it cannot be triggered by an unauthenticated outsider (requires a valid agent session, PR:L), it depends on victim user interaction (viewing the file, UI:P), and it only affects versions before 2.9.0-rc.8 - patched instances normalize the content type and apply a sandbox CSP to attachments.

This is a medium-severity stored XSS (CVSS 4.0 base 5.1) that realistically ranks as a moderate, not top, priority. The CVSS 4.0 vector (AV:N/AC:L/AT:N/PR:L/UI:P/VC:L/VI:L/VA:N with subsequent-system SC:L/SI:L) is internally consistent with the description: network-reachable upload endpoint, low complexity, but requiring an authenticated agent (PR:L) to plant the payload and a victim to view it (UI:P). Impact is capped at Low across confidentiality and integrity because execution is confined to the browser session context rather than the server. No EPSS score, KEV entry, or public PoC was provided in the input; per the available signals this is 'no public exploit identified at time of analysis.' There is no signal conflict here - all sources point to a privilege- and interaction-gated web vulnerability. The main residual risk is privilege escalation within the application: an attacker with a low-privilege agent account can execute script in an administrator's session, which is why this still merits timely patching despite the medium score. Note the PR:L requirement means an external unauthenticated attacker cannot exploit this directly; it presumes a malicious or compromised agent account.

A malicious or compromised low-privilege LibreDesk agent uploads an SVG containing a <script> payload via /api/v1/media, deliberately setting the Content-Type to 'image/svg+xml; x' so it evades the exact-string SVG sandbox check. When another agent or an administrator later opens that media via GET /uploads/{uuid}, the server serves it inline without the sandbox CSP and the script runs in the victim's authenticated session. No public exploit code was provided in the input, but the technique is fully described in the VulnCheck advisory and the fix commit.