render-photo-grid-card
Pass
Audited by Gen Agent Trust Hub on Jul 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The script "scripts/render.py" uses "subprocess.run" to call the "ffmpeg" utility for extracting video frames and encoding the final MP4. The command is executed as a list of arguments without a shell, which prevents common shell injection vulnerabilities.
- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection vulnerability in "scripts/build_card.py" due to improper sanitization of configuration data. Specifically, the "headline" field from the input JSON is directly embedded into the HTML template without escaping, while other fields like "sub" and "wordmark" are correctly handled. This creates an XSS vulnerability within the Playwright-managed browser instance used for rendering. * Ingestion points: Configuration parameters are ingested via a JSON file provided to "build_card.py" and "render.py". * Boundary markers: No explicit instructions are provided to the model or renderer to ignore embedded commands or scripts in the configuration data. * Capability inventory: The skill can read/write files and execute the "ffmpeg" binary via subprocess calls. * Sanitization: While a helper function "esc" (using "html.escape") is implemented and used for most fields, it is omitted for the "headline" parameter, leaving it vulnerable.
Audit Metadata