agent-orchestrator-manager

Parallel task coordination uses the /spawn command to delegate independent work to subagents simultaneously. Each agent handles a separate component—backend, frontend, database—while the orchestrator tracks progress via UUID-based output directories and generates summary reports when complete.

Delegation spawns specialized subagents for each task, enabling true parallelization and isolation of work. Direct execution runs sequentially in a single process. Delegation suits complex, multi-step workflows where independent components can run concurrently under orchestration and supervision.

Yes. Orchestration is designed for full-stack workflows: spawn one agent for backend API development, another for frontend UI, and a third for database schema changes. All run in parallel while the orchestrator tracks completion and aggregates results.

Each spawned agent writes output to a UUID-based directory. The orchestrator monitors these directories to track progress across all parallel tasks, then consolidates findings into a unified summary report upon completion.

Workflows with independent, parallelizable components benefit most: research + planning + implementation running concurrently, or feature work split across multiple system layers. Tasks requiring sequential dependencies are less ideal than those with isolated, concurrent work streams.

No. The Skill operates with no external dependencies or components listed. It uses the /spawn command and UUID-based output directories as its core primitives for delegation and progress tracking across subagents.