What problem does it solve? Running multiple AI agents simultaneously introduces race conditions, file overwrite conflicts, and cascading failures when one agent crashes. This Skill provides orchestration patterns for safely dispatching, executing, and merging work from parallel agent swarms. ## Core Features & Use Cases - Fan-Out/Fan-In Orchestration: Dispatch tasks to multiple worker agents concurrently and synthesize their outputs into a unified result. - Workspace Collision Prevention: Enforce directory isolation and file-level boundaries so concurrent agents never overwrite each other's changes. - Resilient Failure Handling: Use allSettled-style aggregation so one failed agent branch does not crash the entire pipeline. - Use Case: Run a reviewer swarm where four agents audit generated code simultaneously for security, typing, logic, and performance, then merge their reports into a single synthesis for human approval. ## Quick Start Orchestrate three parallel agents to review this codebase for security, performance, and accessibility issues, then merge their findings into one report.
