The Agent Skills Directory

[SAFE]: No malicious patterns, obfuscation, or unauthorized data exfiltration were detected. The skill's behavior is consistent with its stated purpose of architecture review.
[DATA_EXPOSURE]: The skill reads local codebase files, git diffs, and architectural documentation (ADRs). This is restricted to the target path provided by the user and is necessary for its core functionality.
[COMMAND_EXECUTION]: The skill mentions running verification commands after refactoring and using subagents for codebase exploration. These are standard operations within a code analysis environment and are used to ensure the integrity of the refactored code.

architecture-review