infographic-creation
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input and interpolates it directly into a JavaScript string within a generated HTML template.
- Ingestion points: User-supplied text is used to populate the {syntax} and {title} placeholders in the HTML file created during Step 2.
- Boundary markers: The instructions lack guidance for the agent to escape or sanitize user content before embedding it in the file structure.
- Capability inventory: The skill uses a file-writing tool to produce a local HTML file intended for user execution.
- Sanitization: No escaping mechanism is implemented to prevent user-supplied characters (such as backticks or HTML tags) from breaking the JavaScript context or facilitating cross-site scripting (XSS) in the user's browser.
- [EXTERNAL_DOWNLOADS]: The skill generates HTML files that depend on external scripts and assets loaded at runtime.
- The AntV rendering library is loaded from unpkg.com via a script tag.
- The resource loader fetches SVG icons from Iconify and illustrations from a third-party GitHub repository (balazser/undraw-svg-collection).
Audit Metadata