scientific-slides

Pass

Audited by Gen Agent Trust Hub on Jun 13, 2026

Risk Level: SAFE
Full Analysis
  • [DATA_EXFILTRATION]: The skill transmits user-provided slide descriptions and local image files (when using the --attach flag) to the OpenRouter API (https://openrouter.ai/api/v1) for processing. This behavior is consistent with the skill's primary purpose of AI-assisted image generation. OpenRouter is a well-known service for accessing large language models.
  • [COMMAND_EXECUTION]: Several scripts (generate_slide_image.py, generate_schematic.py) use the subprocess.run method to orchestrate internal script calls. Additionally, validate_presentation.py executes the pdflatex utility to verify LaTeX Beamer files. Notably, the developer has included the -no-shell-escape flag in the LaTeX command, which is a critical security measure to prevent malicious LaTeX documents from executing arbitrary system commands.
  • [CREDENTIALS_UNSAFE]: The skill requires an OpenRouter API key. The documentation correctly instructs users to provide this key through the OPENROUTER_API_KEY environment variable. There are no hardcoded secrets or sensitive credentials within the files.
  • [EXTERNAL_DOWNLOADS]: The skill does not download or execute remote code at runtime. It relies on standard, reputable Python libraries (such as requests, Pillow, and PyMuPDF) which are expected for its functionality.
  • [PROMPT_INJECTION]: The skill incorporates extensive system guidelines into its AI prompts to ensure visual quality and scientific standards. No attempts to bypass safety filters or override agent behavior were detected.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 13, 2026, 06:23 AM