The Agent Skills Directory

[SAFE]: The skill is focused on generating technical documentation and does not contain any prompt injection or obfuscation techniques.
[SAFE]: File access is limited to reading repository content and writing documentation files using the allowed read_file and write_file tools. No access to sensitive paths (e.g., .ssh, .aws) is requested.
[SAFE]: No remote code execution or external dependencies are defined. The skill does not perform any network operations.
[SAFE]: Instructions explicitly mandate that changes to existing files must be presented as diffs rather than full replacements, which acts as a safety guard against accidental data loss or unauthorized code modifications.
[SAFE]: Validation steps include verifying the accuracy of commands and paths against the repository, ensuring the documentation reflects the actual system state.

doc-engine