video-to-superprompt
Pass
Audited by Gen Agent Trust Hub on Aug 26, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructions direct the agent to execute shell commands (
ffprobeandffmpeg) using the$VIDEOvariable to analyze media files. While the provided command snippets use quotes around the variable to mitigate simple injection, executing shell commands based on user-supplied paths remains a potential surface for command injection if the environment or input is not strictly validated. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because its core workflow involves analyzing external data, including video files and associated source code (HTML, CSS, and JavaScript). Maliciously crafted instructions embedded in these external assets could attempt to influence the agent's logic during the analysis phase.
- Ingestion points: Video files, metadata, and web source code (HTML/CSS/JS) as described in the
Workflowsection ofSKILL.md. - Boundary markers: Absent. The instructions do not specify the use of delimiters or warnings to ignore instructions found within the analyzed assets.
- Capability inventory: Shell execution (
ffprobe,ffmpeg) and file system access (mkdir, frame extraction). - Sanitization: Absent. There is no explicit instruction to sanitize or filter the ingested code or metadata before processing.
Audit Metadata