The Agent Skills Directory

[SAFE]: The skill is designed for task management and orchestration, defining a clear structure for decomposing projects into parallel or sequential tasks.
[COMMAND_EXECUTION]: The instructions leverage system capabilities such as file access, semantic search, and command execution. These are used strictly within the context of fulfilling user-initiated tasks and managing project workflows.
[INDIRECT_PROMPT_INJECTION]: The skill identifies a potential attack surface by accepting "additional instructions" (追加指示) from users. It mitigates this risk by explicitly instructing the agent to stop and confirm with the user if an instruction contradicts the skill's core safety and orchestration logic. Furthermore, it requires a "Completion Self-Review" to identify unverified assumptions or inaccuracies before concluding a task.

execute