The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Bash and Task tools to orchestrate its workflow. The Task tool is used to spawn internal subagents from the gamedev namespace, and Bash is used for environment checks. These actions are aligned with the stated purpose of coordinating a development team.
[DATA_EXFILTRATION]: Analysis shows no patterns of data exfiltration. While the skill reads local configuration and design files (e.g., design/gdd/, production/review-mode.txt), it does not contain network calls, external URLs, or methods to transmit data to remote servers.
[PROMPT_INJECTION]: The instructions are structured for workflow management and do not contain any patterns intended to bypass safety filters, override system constraints, or extract internal prompts.
[SAFE]: The skill demonstrates best practices for agent safety by implementing mandatory human-in-the-loop (HITL) checkpoints. The AskUserQuestion tool is invoked at every stage gate, ensuring the user reviews and confirms all subagent proposals before the orchestrator proceeds to the next phase or delegates file-writing operations.

team-level