The Agent Skills Directory

[SAFE]: No malicious patterns or security risks were detected in the skill instructions. The process is designed to be highly controlled by the user.
[PROMPT_INJECTION]: The skill does not contain instructions that attempt to bypass safety filters or override system constraints. It uses natural instructional language to define a developer workflow.
[DATA_EXFILTRATION]: No network activity or access to sensitive files (e.g., credentials, environment variables) was detected. The file system operations are limited to managing plan files in a dedicated directory.
[REMOTE_CODE_EXECUTION]: The skill does not download or execute external code. It refers to other internal agent skills for testing and refactoring, which is standard behavior for agent-based tasks.
[COMMAND_EXECUTION]: All git commit operations require explicit user approval, which serves as a critical security checkpoint against unintended code changes.

planning