Jellyfin CVE-2026-35033

Jellyfin versions prior to 10.11.7. Remote unauthenticated attacker crafts malicious StreamOptions query parameter with ffmpeg drawtext filter injection. No victim interaction required. Default streaming endpoint vulnerable.

This vulnerability presents critical real-world risk despite requiring item GUID knowledge. The CVSS 4.0 score of 9.3 reflects network accessibility (AV:N), low attack complexity (AC:L), no authentication (PR:N), and high confidentiality impact (VC:H, VI:H) with subsequent system impact (SC:L, SI:L). While no public exploit is identified and EPSS data is unavailable for this recent CVE, the detailed technical disclosure in the GitHub Security Advisory provides a complete exploitation blueprint including specific attack payloads. The SSVC decision framework indicates high technical impact (arbitrary file read enables credential theft, configuration exposure, and lateral movement preparation) with medium exploitation barriers-attackers need valid item GUIDs typically obtained through authenticated reconnaissance, though brute-forcing GUIDs or discovering them in public Jellyfin instances remains possible. The authentication bypass tag and unauthenticated access to the vulnerable endpoint elevate urgency. Organizations running internet-facing Jellyfin instances below version 10.11.7 face immediate risk of sensitive data exfiltration. The vulnerability's weaponization potential is high given the popularity of Jellyfin in home lab and small business environments where security monitoring may be limited.

An unauthenticated attacker discovers a publicly accessible Jellyfin server through Shodan or similar scanning. By enumerating or guessing valid media item GUIDs (possibly leaked in shared library links or brute-forced), the attacker crafts a malicious HTTP request to /Videos/{validItemGuid}/stream with an injected query parameter like '?drawtext=textfile=/etc/shadow:fontsize=48'. The vulnerable ParseStreamOptions method concatenates this directly into the ffmpeg command line, causing ffmpeg to render the contents of /etc/shadow as text overlay in the returned video stream. The attacker downloads the video and extracts credentials, database connection strings, API keys from configuration files, or SSH keys to escalate access to the underlying server infrastructure.