The Agent Skills Directory

[SAFE]: The skill serves as a meta-utility for skill development, providing guidelines and automation scripts for repository maintenance.
[COMMAND_EXECUTION]: The skill includes Python scripts (quick_validate.py, init_skill.py, package_skill.py, fix_mojibake.py, split_oversized_skills.py, and upgrade_dual_compat.py) designed for repository management. These scripts perform standard operations such as file reading/writing, YAML parsing, and creating ZIP archives. They do not invoke sensitive shell commands or access unauthorized system resources.
[EXTERNAL_DOWNLOADS]: The skill does not perform any automated downloads. While the documentation in references/skill-authoring-best-practices.md provides examples of package installation commands (e.g., pip install pypdf), these are instructional guidelines for developers and are not executed by the skill itself.
[DATA_EXFILTRATION]: No network activity or exfiltration patterns were detected. All scripts operate exclusively on local files within the project directory.
[PROMPT_INJECTION]: The skill instructions specifically promote safety and security, directing authors to create skills that are "Secure by default" and "Explicit about failure handling." No adversarial injection patterns were found.

skill-writing