The Agent Skills Directory

[SAFE]: The skill defines a complex multi-agent orchestration workflow. The 'hostile' and 'adversarial' role-play instructions are scoped to a red-teaming planning methodology intended to identify over-engineering and weak assumptions. It does not contain instructions to bypass AI safety guardrails, exfiltrate data, or execute unauthorized commands. The use of platform-specific orchestration tools (team_*) and the task tool for plan formalization is consistent with the skill's stated purpose.

hyperplan