The Agent Skills Directory

[SAFE]: No malicious patterns or security vulnerabilities were identified. The skill operates as a deterministic gate and does not use LLM invocation for decision-making.
[DATA_EXFILTRATION]: The skill accesses local files to perform audits. It implements strict path validation using Path.resolve() and is_relative_to() to ensure it only reads files within the active workspace. Additionally, it restricts file access to the .tex extension and imposes a 5MB size limit to prevent unauthorized access to sensitive system files or denial-of-service via large files.
[COMMAND_EXECUTION]: The entrypoint executes a local Python script using a fixed command line. No arbitrary command execution or subprocess spawning based on user input was detected.

paper-length-gate