The Agent Skills Directory

[SAFE]: The skill provides structural templates and automation for developers. The primary operations are local file creation and text analysis within the user's own repository.
[COMMAND_EXECUTION]: The skill includes two Bash scripts, new-skill.sh and lint-skill.sh.
new-skill.sh handles skill name input with strict sanitization ([a-zA-Z0-9_-]), effectively preventing command injection or path traversal when creating directories or copying templates.
lint-skill.sh performs static analysis of Markdown and script files using standard tools like awk, sed, and grep without executing the files being analyzed.
[REMOTE_CODE_EXECUTION]: The documentation references external tools like uvx and npx and provides templates for PEP 723 Python scripts. These are presented as best-practice examples for the user to implement in their own skills rather than being executed by the skill-author skill itself. No automatic remote code execution was detected.
[DATA_EXFILTRATION]: There are no network operations, credential accesses, or sensitive data reads performed by the skill's scripts or instructions.
[PROMPT_INJECTION]: The skill instructions are purely instructional and educational, with no attempts to override system prompts or bypass safety filters.

skill-author