article-magazine

Pass

Audited by Gen Agent Trust Hub on Sep 10, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest user-supplied Markdown or notes and transform them into a polished HTML layout, which introduces a surface for indirect prompt injection if the source data contains malicious instructions.
  • Ingestion points: The example_prompt in SKILL.md takes arbitrary user content to be processed by the agent into the magazine layout.
  • Boundary markers: The template instructions do not specify the use of delimiters or 'ignore embedded instructions' warnings when handling user-provided notes.
  • Capability inventory: The skill generates HTML with embedded styles and references external scripts/fonts, which could be leveraged if the agent obeys instructions hidden in the input text.
  • Sanitization: No sanitization or validation logic is defined for the input content prior to its interpolation into the HTML template.
  • [EXTERNAL_DOWNLOADS]: The example article fetches styling and font assets from well-known technology services.
  • Loads the Tailwind CSS framework from https://cdn.tailwindcss.com in example.html.
  • Fetches typography from Google Fonts via fonts.googleapis.com and fonts.gstatic.com in example.html.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 10, 2026, 08:00 PM