The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes the xelatex and bibtex tools via the Python subprocess module to verify that generated LaTeX content compiles correctly. The commands are constructed using a fixed list of arguments, and the compiler engine name is retrieved from a configuration file.
[INDIRECT_PROMPT_INJECTION]: The skill processes project assets (LaTeX files, BibTeX entries, and source code) to provide context for the AI generation process. To mitigate the risk of malicious data influencing the agent, the SecurityManager class performs post-generation scanning and sanitization, filtering out a blacklist of potentially dangerous LaTeX keywords and commands.
[DATA_EXFILTRATION]: The skill provides a mechanism for reporting bugs and design flaws to the author's specific GitHub repository (huangwb8/bensz-bugs) using the gh CLI tool. This functionality is intended for error reporting and is only activated upon explicit user request.
[SAFE_PRACTICE]: The skill implements a robust security framework that includes mandatory file backups before modification, path normalization to prevent directory traversal, and strict access controls that prevent the modification of critical system files like main.tex and @config.tex.

complete-example