The Agent Skills Directory

[SAFE]: The skill is restricted to text generation for commit messages. It does not perform network operations, access sensitive files, or execute shell commands.
[SAFE]: The instructions include clear functional boundaries, explicitly stating that the skill does not stage, commit, or amend files, ensuring that final actions require manual user intervention.
[SAFE]: No obfuscation techniques, hidden URLs, or malicious instructional patterns were found in the skill or its associated documentation.

caveman-commit