The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses subprocess.run to call external tools such as pandoc, xelatex, latexmk, and soffice (LibreOffice) for document conversion and compilation. These operations are essential to the skill's functionality. Command construction uses argument lists rather than shell strings, which effectively prevents shell injection vulnerabilities.
[INDIRECT_PROMPT_INJECTION]: As the skill processes user-provided documents (e.g., contest questions and datasets), it has a potential surface for indirect prompt injection. To mitigate this, the skill implements a structured 'Quality Inspection' workflow. It provides explicit system instructions for subagents, defining them as read-only inspectors and establishing clear boundaries for their role.
[EXTERNAL_DOWNLOADS]: The skill performs academic searches by querying the OpenAlex and AnySearch APIs. These are established, legitimate services used for retrieving scholarly metadata and abstracts. No evidence of unauthorized data exfiltration or execution of remote untrusted scripts was found.

math-modeling