lower-thirds
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The
scripts/render.mjsscript automates the video creation process by callingffmpegandffprobe. These system calls are implemented using argument arrays inspawnandexecFileSync, which prevents shell-based command injection by not invoking a shell to parse the input. - [EXTERNAL_DOWNLOADS]: The skill requires the
playwright-corelibrary to control a browser instance and downloads font assets from Google Fonts' official service (fonts.googleapis.com). These dependencies and references target well-known services and are used for their intended design purposes. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process user-provided strings (such as names and roles) to generate graphic overlays.
- Ingestion points: User data is placed in the
CARDSconstant within the HTML template. - Boundary markers: None specific, but the data is confined to a JavaScript object.
- Capability inventory: File writing (videos/images) and command execution (ffmpeg) are used by the rendering script.
- Sanitization: The user-provided text is rendered exclusively as visual elements on an HTML5 Canvas using
ctx.fillText, which treats the data as literal strings rather than executable code.
Audit Metadata