The Agent Skills Directory

[SAFE]: The skill implements a multi-agent orchestration pattern designed to improve task efficiency through parallel execution and structured handoffs.
[SAFE]: Tool access is restricted to standard development utilities (Read, Write, Edit, Bash, Glob, Grep) used for project maintenance and code analysis.
[SAFE]: The memory protocol is limited to local project paths (.claude/context/memory/) for persistent state management, which is a common practice for maintaining context across agent sessions.
[SAFE]: Execution scripts and hooks are minimal, serving primarily as scaffolding for metadata and basic validation without executing dangerous shell commands or downloading external content.
[SAFE]: The skill's instructions focus on operational efficiency and error handling, such as limiting the number of parallel workers to prevent coordination overhead.

swarm-coordination