The Agent Skills Directory

[COMMAND_EXECUTION]: The script executes external LaTeX binaries (latexmk, xelatex, bibtex, pdfinfo) to process files within the workspace. It uses a secure implementation by passing arguments as a list to subprocess.run in scripts/run.py, which prevents shell-based command injection.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted LaTeX source files and bibliography data provided in the workspace. While this creates a potential surface for indirect prompt injection if the AI agent interprets instructions embedded in error logs or source snippets, this is inherent to the skill's primary purpose of processing and reporting on user-provided content.
Ingestion points: LaTeX source (latex/main.tex) and bibliography (citations/ref.bib) files are read from the workspace as defined in scripts/run.py.
Boundary markers: None identified; the files are passed directly to the toolchain.
Capability inventory: The skill has the ability to execute local command-line tools and perform file system operations (write reports, copy PDFs) within the provided workspace.
Sanitization: The skill does not sanitize the contents of the TeX files or the resulting build logs before they are presented in the markdown report.

latex-compile-qa