latex-paper-conversion
Pass
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: SAFEDYNAMIC_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [DYNAMIC_EXECUTION]: The skill instructs the agent to generate and execute a Python script (e.g.,
convert_format.py) at runtime to parse, extract, and map content between different LaTeX templates. - [COMMAND_EXECUTION]: The skill executes various system commands to perform its core functions, including LaTeX compilers (
pdflatex), bibliography tools (bibtex,biber), and search utilities (grep,rg) for log analysis. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user-provided LaTeX files, which represent an attack surface for embedded malicious instructions targeting the agent or the LaTeX compiler.
- Ingestion points: User-supplied
.texsource files and template directories (SKILL.md). - Boundary markers: The skill mentions "boundary-aware extraction" but does not specify fixed delimiters for prompt interpolation.
- Capability inventory: The skill has access to shell command execution (compilers) and Python script execution (SKILL.md).
- Sanitization: The instructions include a critical security control by mandating that compilation occur "without shell escape" to prevent the TeX engine from executing arbitrary system commands.
Audit Metadata