latex-posters

Pass

Audited by Gen Agent Trust Hub on Sep 25, 2026

Risk Level: SAFEDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [DYNAMIC_EXECUTION]: The script scripts/generate_schematic.py uses subprocess.run to execute a local helper script (generate_schematic_ai.py). This allows for modular tool execution but introduces a surface for dynamic process creation.
  • [COMMAND_EXECUTION]: The skill includes scripts/review_poster.sh, a bash utility that executes external commands such as pdfinfo, pdffonts, and pdfimages to validate PDF properties and quality.
  • [INDIRECT_PROMPT_INJECTION]: The skill generates visual content by passing user-supplied descriptions to external LLMs (Gemini models via OpenRouter).
  • Ingestion points: The natural language prompt argument passed to the schematic generation scripts.
  • Boundary markers: Absent; user input is interpolated directly into message templates for the LLM.
  • Capability inventory: File system write access for images, local script execution, and outbound network requests to API endpoints.
  • Sanitization: No explicit escaping or filtering is applied to user-provided strings before they are processed by the LLM.
  • [EXTERNAL_DOWNLOADS]: The scripts/generate_schematic_ai.py script communicates with openrouter.ai via the requests library to fetch AI-generated images and critique results, representing interaction with a well-known external service.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 25, 2026, 02:34 PM