render-value-prop
Pass
Audited by Gen Agent Trust Hub on Jul 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill orchestrates a video rendering pipeline using subprocess.run to invoke FFmpeg and internal helper scripts like render_hyperframe.py. These calls are used to process video frames and concatenate them into a final MP4 file, which is the primary purpose of the skill.
- [EXTERNAL_DOWNLOADS]: HTML templates and CSS files reference external font resources from api.fontshare.com and fonts.googleapis.com. These are well-known services commonly used for web-based rendering and do not pose a security risk in this context.
- [PROMPT_INJECTION]: The skill exhibits an indirect injection surface by interpolating values from the config.json file directly into HTML templates without sanitization. This content is subsequently rendered in a headless browser environment.
- Ingestion points: Data from config.json (such as headline_html, label, and benefit_sentence) is read and processed in scripts/render_master.py and other script files.
- Boundary markers: No delimiters or 'ignore' instructions are used when embedding configuration values into the HTML templates.
- Capability inventory: The rendering process uses Playwright, which can execute JavaScript and access the local filesystem via the file:// protocol. The skill also invokes subprocesses for FFmpeg and Python orchestration.
- Sanitization: The skill does not perform HTML escaping or validation on the strings provided in the configuration before rendering them.
Audit Metadata