skills/pietz/skills/slides/Gen Agent Trust Hub

slides

Pass

Audited by Gen Agent Trust Hub on Sep 21, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes user-generated or external HTML content through a headless Chromium browser to generate PDFs. This creates a surface where malicious JavaScript embedded in the HTML could attempt to exfiltrate information or perform unauthorized network requests during the rendering process.
  • Ingestion points: The input argument in scripts/html_to_pdf.py receives the HTML file path for processing.
  • Boundary markers: There are no explicit delimiters or instructions provided to the agent or the rendering engine to ignore script execution within the HTML content.
  • Capability inventory: The skill utilizes Playwright, which provides full browser capabilities, including network access and the ability to access the local file system via file:// URIs.
  • Sanitization: The Python script does not perform any sanitization or filtering of the HTML content before passing it to the Playwright engine.
  • [COMMAND_EXECUTION]: The script scripts/html_to_pdf.py uses subprocess.run to ensure that the Chromium browser binaries required by Playwright are installed. The command playwright install chromium is hardcoded and executes via the current Python executable.
  • [EXTERNAL_DOWNLOADS]: The skill instructions and sample code reference several well-known content delivery networks (CDNs) for assets:
  • Fetches Tailwind CSS from tailwindcss.com.
  • Loads typography from Google Fonts (fonts.googleapis.com and fonts.gstatic.com).
  • Downloads iconography from Lucide via unpkg.com.
  • The script also triggers a download of Chromium binaries from the official Playwright registry during the first run.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 21, 2026, 12:13 PM