The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes multiple shell commands through the Bash tool, including LaTeX build utilities (latexmk, pdflatex) and a local Python verification script located at $HOME/.claude/skills/_shared/verify_outputs.py.
[PROMPT_INJECTION]: The skill presents a surface for indirect prompt injection (Category 8) by processing untrusted data to perform file modifications.
Ingestion points: The skill reads and parses content from .tex source files and out/*.log build logs found in the user's project directories.
Boundary markers: There are no boundary markers or specific instructions to the agent to disregard malicious patterns or instructions embedded within the processed LaTeX files or logs.
Capability inventory: The skill possesses Edit, Write, and Bash capabilities, which are used to apply "auto-fixes" (such as adding packages) based on the content parsed from the logs.
Sanitization: No sanitization or validation logic is defined to check the integrity of the error patterns or package names extracted from the logs before they are used to edit source files.
[COMMAND_EXECUTION]: The skill accesses the local configuration file at ~/.config/task-mgmt/research-root to determine the project root directory. This path is consistent with the author's namespace and represents expected vendor functionality.

latex-health-check