architecture-diagram
Pass
Audited by Gen Agent Trust Hub on Apr 27, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection via untrusted user input used to build HTML diagrams.\n
- Ingestion points: User descriptions for components and labels used in the SVG generation logic within SKILL.md.\n
- Boundary markers: No specific delimiters or markers are defined to wrap or separate user-supplied input from the static HTML template.\n
- Capability inventory: Uses the
write_filetool to save the generated HTML file to the user's filesystem.\n - Sanitization: No instructions are provided for escaping or sanitizing user input strings (e.g., to prevent the inclusion of tags) before they are written to the HTML file.\n- [EXTERNAL_DOWNLOADS]: The provided template references typography from a well-known external service.\n
- The
templates/template.htmlfile includes a stylesheet link to Google Fonts to load the 'JetBrains Mono' font family. This is a neutral reference to a standard service and does not trigger verdict escalation.
Audit Metadata