quarto-doc-setup

Fail

Audited by Gen Agent Trust Hub on Jun 13, 2026

Risk Level: HIGHCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes several shell commands to manage the environment and project workflow. Notably, it instructs the agent to use sudo tlmgr install for installing LaTeX packages, which involves executing commands with elevated administrative privileges. Other executed commands include quarto, pdflatex, kpsewhich, fc-list, and brew.
  • [EXTERNAL_DOWNLOADS]: The skill automates the download of the 'JetBrains Mono' font using brew install --cask. It also references quarto.org as a source for software installation.
  • [PROMPT_INJECTION]: The skill demonstrates an indirect injection surface when processing document metadata.
  • Ingestion points: User-provided Title, Author, Organization, and other fields defined in SKILL.md (Step 3).
  • Boundary markers: Metadata fields are wrapped in double quotes within the YAML frontmatter during the creation of the .qmd file.
  • Capability inventory: The skill uses quarto render, which spawns subprocesses for LaTeX compilation and involves file system write operations.
  • Sanitization: No explicit sanitization or validation of user input is performed before interpolation into the generated project files.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Jun 13, 2026, 01:04 AM
Security Audit — agent-trust-hub — quarto-doc-setup