Executive brief
ZLMediaKit is a streaming media server that exposes an HTTP API for configuration and control. An unauthenticated attacker can inject arbitrary shell commands through the setServerConfig API endpoint by overwriting the ffmpeg.snap configuration parameter, then execute those commands by triggering the getSnap endpoint, gaining code execution as the ZLMediaKit process user (typically root). Versions prior to February 2025 ship with a well-known default API secret, making exploitation trivial on most deployed systems.
Technical details
The vulnerability is an OS command injection in FFmpegSource.cpp where the ffmpeg.snap configuration value is used directly in a snprintf() call to build a shell command that is then executed via Process::run() and execv() without any validation or sanitization. The setServerConfig API endpoint accepts arbitrary key-value pairs and writes them to the in-memory configuration without checking for shell metacharacters. An attacker with knowledge of the API secret (default: 035c73f7-bb6b-4889-a715-d9eb2d1925cc on pre-2025-02-21 versions, or obtainable through information disclosure) can inject commands that execute with the privileges of the ZLMediaKit process.
Affected products
- ZLMediaKit ZLMediaKit all versions
Timeline
- 2026-08-07: disclosed: GitHub security advisory GHSA-qh3r-f727-ww9m published
- 2025-02-21: other: Partial mitigation: auto-randomization of API secret in commit aa160c6, but injection vector remains unfixed