The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/doc_to_docx.py script uses subprocess.run to invoke LibreOffice for document conversion. The implementation uses a list-based argument structure and avoids shell execution, effectively mitigating the risk of command injection.\n- [EXTERNAL_DOWNLOADS]: The setup.sh script manages the installation of required Python packages and the LibreOffice suite. It uses standard package managers (pip, brew, apt, dnf, pacman) and includes mandatory user confirmation prompts before proceeding with large installations.\n- [PROMPT_INJECTION]: The SKILL.md file contains instructional guidelines for 'Anti-Hallucination' and 'Core Principles' to ensure document fidelity. These instructions are benign and do not attempt to override the agent's safety protocols or core behavior.\n- [DATA_EXFILTRATION]: Analysis of the provided scripts and markdown documentation shows no evidence of network communication to untrusted domains or attempts to harvest sensitive files such as SSH keys or credentials. The skill's operations are limited to the local filesystem for processing user-provided templates.

fill-protocol