video-edit
Pass
Audited by Gen Agent Trust Hub on Nov 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONPRIVILEGE_ESCALATION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes shell commands including
ffmpeg,ffprobe,printf, andrmfor video processing, file system output, and cleanup operations. - [PRIVILEGE_ESCALATION]: Setup instructions for Linux environments specify the use of
sudofor package installation viaapt, which grants administrative privileges to the installation process. - [INDIRECT_PROMPT_INJECTION]: The agent is directed to ingest and process metadata from external video files, creating a vulnerability surface for adversarial instructions embedded in file tags.
- Ingestion points: Technical metadata and stream information extracted by
ffprobein JSON format from user-supplied video files. - Boundary markers: Not present; the instructions provide no delimiters or warnings to the agent to treat metadata as untrusted or to ignore embedded natural language.
- Capability inventory: The skill enables execution of
ffmpegwith file overwrite permissions (-y) and file deletion viarm. - Sanitization: Metadata values are not filtered, validated, or sanitized before being returned to the agent context.
Audit Metadata