skills/mathbullet/skills/html/Gen Agent Trust Hub

html

Pass

Audited by Gen Agent Trust Hub on Sep 4, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill configures generated HTML documents to fetch resources from well-known and trusted external services. These include typography assets from Google Fonts (fonts.googleapis.com and fonts.gstatic.com), the MathJax library from cdn.jsdelivr.net for mathematical formulas, and Highlight.js from cdnjs.cloudflare.com for code syntax highlighting.\n- [COMMAND_EXECUTION]: The skill includes a bash utility script, render-pdf.sh, designed for macOS. The script identifies local installations of Google Chrome or Chromium and executes them in headless mode using the --print-to-pdf flag to convert HTML files. The script implements security best practices such as shell quoting and set -euo pipefail.\n- [INDIRECT_PROMPT_INJECTION]: The skill acts as a transformer for user-provided technical text into HTML format, which introduces an indirect injection surface.\n
  • Ingestion points: Technical descriptions and mechanism explanations provided to the agent as described in SKILL.md.\n
  • Boundary markers: None explicitly defined in the instructions to separate untrusted user input from HTML templates.\n
  • Capability inventory: The skill can write files to the local system and execute a headless browser via the included PDF rendering script.\n
  • Sanitization: No explicit sanitization or escaping steps are defined for the input text, which could allow for the inclusion of scripts that execute when the resulting HTML is viewed or rendered.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 4, 2026, 09:18 PM
Security Audit — agent-trust-hub — html