video-processing-editing
Warn
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: MEDIUMDYNAMIC_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [DYNAMIC_EXECUTION]: The script
scripts/video_editor.pyuses the Pythoneval()function to process ther_frame_ratemetadata field returned byffprobe. Since this metadata originates from the input video file, a maliciously crafted file with a forged frame rate field could potentially execute arbitrary Python code when the metadata is extracted and evaluated. - [COMMAND_EXECUTION]: The skill extensively uses
subprocess.run()across multiple files (scripts/video_editor.py,scripts/audio_mixer.py,scripts/batch_processor.py,scripts/quality_assessment.py,scripts/thumbnail_generator.py,scripts/motion_graphics.py,scripts/timelapse_creator.py) to execute FFmpeg and ffprobe commands. While this is the primary purpose of the skill, the execution of shell commands on user-provided files presents a significant capability surface. - [INDIRECT_PROMPT_INJECTION]: The skill is vulnerable to indirect instruction injection through the processing of untrusted external data.
- Ingestion points: External video, audio, and image files provided to the scripts as inputs (e.g., in
video_editor.py,batch_processor.py). - Boundary markers: No explicit boundary markers or warnings are used to instruct the agent to ignore embedded metadata or structural data within the processed files.
- Capability inventory: The skill has the capability to execute shell commands (via FFmpeg/ffprobe) and run arbitrary Python code (via
eval()invideo_editor.py). - Sanitization: While most subprocess calls use list-based arguments, the use of
eval()on external metadata inscripts/video_editor.pybypasses safe handling and creates an exploitable path for malicious content embedded in file headers.
Audit Metadata