The Agent Skills Directory

[SAFE]: The skill is designed for repository maintenance tasks. It operates locally and does not perform any network requests or external data exfiltration.
[COMMAND_EXECUTION]: The skill executes git, python3, and node commands to perform its functions. These executions use safe subprocess patterns (passing arguments as lists rather than shell strings) and are limited to the repository context.
[SAFE]: Analysis of the included Python scripts shows they use standard libraries and safe parsing techniques for handling repository configuration and changeset files. No dynamic code execution (e.g., eval/exec) or obfuscation was detected.

changeset-pr