video-frames
Pass
Audited by Gen Agent Trust Hub on Jul 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill invokes
ffmpegandmkdirwithin thescripts/frame.shscript. It correctly utilizes shell quoting for all variable expansions, effectively mitigating standard shell command injection risks. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) due to the handling of external parameters. (1) Ingestion points: The script accepts user-influenced parameters such as
--timeand--indexvia command-line arguments. (2) Boundary markers: Absent; the script does not verify the structure or content of the provided arguments before use. (3) Capability inventory: The script performs system command execution usingffmpeg. (4) Sanitization: There is no sanitization or validation logic for the--indexor--timeinputs, which are interpolated directly into theffmpegfilter string. This could allow an attacker to inject additionalffmpegfilters or options by providing specially crafted strings.
Audit Metadata