The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local initialization script ${CLAUDE_PLUGIN_ROOT}/scripts/cli init during the setup phase to create or verify project configuration files. This appears to be a standard integration with the host environment's CLI tools.
[PROMPT_INJECTION]: The skill functions as an orchestration layer that processes user-provided tasks to define agent roles and assignments. While it lacks explicit boundary markers for this untrusted input, the behavior is consistent with its primary purpose of task analysis and delegation.
[SAFE]: All orchestration capabilities, such as creating teams and spawning agents, utilize built-in platform tools (TeamCreate, Agent, TaskCreate). The skill does not perform any external network operations or access sensitive system files outside of project-specific metadata (e.g., CLAUDE.md).

team-lead