The Agent Skills Directory

[SAFE]: The skill serves as an orchestration framework for multi-agent workflows, utilizing platform-specific tools (Task, Teammate) and state management paths (~/.claude/teams/, ~/.claude/tasks/) that are appropriate and necessary for its stated purpose.
[SAFE]: No evidence of malicious prompt injection, data exfiltration, or obfuscation was found. The instructional content focus is on coordination efficiency, anti-sycophancy patterns, and swarm resilience.
[SAFE]: The skill explicitly promotes safety best practices, including the 'one-owner-per-file' rule to prevent race conditions and the use of 'isolation: worktree' to ensure parallel agents do not corrupt the shared repository state.
[SAFE]: References to external utilities like tmux and the it2 CLI (for iTerm2) are documented as legitimate execution backends. Installation instructions provided (e.g., uv tool install it2) use well-known, trusted package managers.
[SAFE]: The skill implements structured handoff protocols and review gates (Stage 1: Spec Compliance, Stage 2: Code Quality) which act as internal security and quality controls for generated content.

orchestrating-swarms