skills/ses2905/ai-toolkit/slides/Gen Agent Trust Hub

slides

Pass

Audited by Gen Agent Trust Hub on Sep 20, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill fetches design assets and font pairings from well-known services, specifically Google Fonts (fonts.googleapis.com) and Fontshare (api.fontshare.com), to provide typography for the generated slides.
  • [COMMAND_EXECUTION]: The agent is instructed to execute local Python scripts (scripts/search.py) to query a design token database based on user input for presentation 'vibes'.
  • [COMMAND_EXECUTION]: Instructions provide for the execution of system commands such as open (macOS) or start (Windows) to automatically display the resulting HTML presentation files in a browser for the user.
  • [COMMAND_EXECUTION]: The documentation includes suggested commands for using external tools like the vercel CLI and Google Chrome in headless mode to assist with sharing and exporting the final product.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests user-supplied text to populate slide content and uses instructions from reference files to build complex HTML/JS outputs, creating a standard vulnerability surface for generators.
  • Ingestion points: User input gathered during the Content and Style Discovery phases described in SKILL.md.
  • Boundary markers: Not implemented in the templates or instructions.
  • Capability inventory: Includes file system writing, execution of local scripts, and system shell command execution for file opening and deployment.
  • Sanitization: The skill does not perform explicit sanitization of user-provided content before inserting it into the generated presentation file.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 20, 2026, 05:59 AM